diff --git a/.gitignore b/.gitignore index bf36af6..39f1558 100644 --- a/.gitignore +++ b/.gitignore @@ -133,3 +133,4 @@ dmypy.json .pre-commit-config.yaml experimentation/.myconfig experimentation/.tunnel +results diff --git a/analysis_mysql.py b/analysis_mysql.py index b787118..d3d1076 100644 --- a/analysis_mysql.py +++ b/analysis_mysql.py @@ -1,18 +1,40 @@ +import argparse +from typing import Tuple from experimentation.Sets import Datasets from experimentation.Utils import TextColor from experimentation.Database import MySQL -models = ["stree", "odte", "adaBoost", "bagging"] +models = ["stree", "oc1", "cart", "odte", "adaBoost", "bagging"] title = "Best model results" -lengths = (30, 9, 11, 11, 11, 11) +lengths = (30, 9, 11, 11, 11, 11, 11, 11) -def report_header_content(title): +def parse_arguments() -> Tuple[str, str, str, bool, bool]: + ap = argparse.ArgumentParser() + ap.add_argument( + "-e", + "--experiment", + type=str, + choices=["any", "gridsearch", "crossval"], + required=False, + default="any", + ) + args = ap.parse_args() + return args.experiment + + +def report_header_content(title, experiment): length = sum(lengths) + len(lengths) - 1 output = "\n" + "*" * length + "\n" - num = (length - len(title) - 2) // 2 - num2 = length - len(title) - 2 - 2 * num - output += "*" + " " * num + title + " " * (num + num2) + "*\n" + num = (length - len(title) - len(experiment) - 2) // 2 + num2 = length - len(title) - len(experiment) - 5 - 2 * num + output += ( + "*" + + " " * num + + f"{title} - {experiment}" + + " " * (num + num2) + + "*\n" + ) output += "*" * length + "\n\n" lines = "" for item, data in enumerate(fields): @@ -22,8 +44,12 @@ def report_header_content(title): return output -def report_header(exclude_params): - print(TextColor.HEADER + report_header_content(title) + TextColor.ENDC) +def report_header(title, experiment): + print( + TextColor.HEADER + + report_header_content(title, experiment) + + TextColor.ENDC + ) def report_line(line): @@ -59,13 +85,14 @@ def report_footer(agg): ) +experiment = parse_arguments() dbh = MySQL() database = dbh.get_connection() dt = Datasets(False, False, "tanveer") fields = ("Dataset", "Reference") for model in models: fields += (f"{model}",) -report_header(title) +report_header(title, experiment) color = TextColor.LINE1 agg = {} for item in [ @@ -78,15 +105,16 @@ for item in [ agg[item]["worse"] = 0 for dataset in dt: find_one = False + # Look for max accuracy for any given dataset line = {"dataset": color + dataset[0]} - record = dbh.find_best(dataset[0], "any") + record = dbh.find_best(dataset[0], "any", experiment) max_accuracy = 0.0 if record is None else record[5] for model in models: - record = dbh.find_best(dataset[0], model) + record = dbh.find_best(dataset[0], model, experiment) if record is None: line[model] = color + "-" * 9 + " " else: - reference = record[10] + reference = record[13] accuracy = record[5] find_one = True agg[model]["items"] += 1 diff --git a/data/csv/balance-scale.csv b/data/csv/balance-scale.csv new file mode 100755 index 0000000..145846e --- /dev/null +++ b/data/csv/balance-scale.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:175ffaf1d0917b6bc68cdd1dd1ebab53255a024f8b4396a68d1100285a765205 +size 21250 diff --git a/data/csv/balloons.csv b/data/csv/balloons.csv new file mode 100755 index 0000000..cd18998 --- /dev/null +++ b/data/csv/balloons.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea37ce5dc61fa7ae8b86a43ffd37fc8c02bfc0146939b8c7e3cbb5d557773051 +size 640 diff --git a/data/csv/breast-cancer-wisc-diag.csv b/data/csv/breast-cancer-wisc-diag.csv new file mode 100755 index 0000000..5cb49ec --- /dev/null +++ b/data/csv/breast-cancer-wisc-diag.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:919649981eee9f3c33baa29d1e42f25c793a334663a8bafb6503418f1b6961d9 +size 160513 diff --git a/data/csv/breast-cancer-wisc-prog.csv b/data/csv/breast-cancer-wisc-prog.csv new file mode 100755 index 0000000..b5d2426 --- /dev/null +++ b/data/csv/breast-cancer-wisc-prog.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2a09ae7988b245cd356e9aeca073c436af5cf33b1a7133f726321b637bc2bd0 +size 61098 diff --git a/data/csv/breast-cancer-wisc.csv b/data/csv/breast-cancer-wisc.csv new file mode 100755 index 0000000..f21bcca --- /dev/null +++ b/data/csv/breast-cancer-wisc.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:505191e7646eea4b0a6753ee4deff3c67af94cefde906e4677a9e820295ffd19 +size 61013 diff --git a/data/csv/breast-cancer.csv b/data/csv/breast-cancer.csv new file mode 100755 index 0000000..fc2f0d4 --- /dev/null +++ b/data/csv/breast-cancer.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97e18173f275b9a973bcf1af32ccfaaf8ab535d16249f5327a96e5993e683ccc +size 24263 diff --git a/data/csv/cardiotocography-10clases.csv b/data/csv/cardiotocography-10clases.csv new file mode 100755 index 0000000..a1fa19b --- /dev/null +++ b/data/csv/cardiotocography-10clases.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9f6697cb5425eb369312805bd71346a62e569181857c5e27dff0e90840c28b9 +size 424345 diff --git a/data/csv/cardiotocography-3clases.csv b/data/csv/cardiotocography-3clases.csv new file mode 100755 index 0000000..a2b40da --- /dev/null +++ b/data/csv/cardiotocography-3clases.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8aff76f947ea5dc72b533c87f8e8a36c4928003f80efa1e98ba6388f348ebab +size 424345 diff --git a/data/csv/conn-bench-sonar-mines-rocks.csv b/data/csv/conn-bench-sonar-mines-rocks.csv new file mode 100755 index 0000000..43361dc --- /dev/null +++ b/data/csv/conn-bench-sonar-mines-rocks.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a318c6833ffc174d372b36adfa818046589f3edaed41ec682b1a2aee7a57b54a +size 115907 diff --git a/data/csv/cylinder-bands.csv b/data/csv/cylinder-bands.csv new file mode 100755 index 0000000..0064c87 --- /dev/null +++ b/data/csv/cylinder-bands.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5dfe96038e887bd50a957562ae14675f26c4ac4b6d2ffe5051b8c3f9e3b10dd +size 166537 diff --git a/data/csv/dermatology.csv b/data/csv/dermatology.csv new file mode 100755 index 0000000..415af06 --- /dev/null +++ b/data/csv/dermatology.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09a71712d42f751f17399fd3490a32c3fac0a9bcc8f9d53c0aad47027c60e5d9 +size 118348 diff --git a/data/csv/echocardiogram.csv b/data/csv/echocardiogram.csv new file mode 100755 index 0000000..f5f1904 --- /dev/null +++ b/data/csv/echocardiogram.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e4cc0d12968eb63dd5f1ad4cd7f14fa5657d799e041a32fce076741511d8707 +size 12267 diff --git a/data/csv/fertility.csv b/data/csv/fertility.csv new file mode 100755 index 0000000..705c010 --- /dev/null +++ b/data/csv/fertility.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cba1b704b86b3424c4c856d042b894da7d68b6245f04ac44a7ef2c31492df53b +size 8392 diff --git a/data/csv/haberman-survival.csv b/data/csv/haberman-survival.csv new file mode 100755 index 0000000..0af9725 --- /dev/null +++ b/data/csv/haberman-survival.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba7abb57fe3819b3c4312bf22c25edc6af2851d825ee593586eeac0bd6e6cb61 +size 9138 diff --git a/data/csv/heart-hungarian.csv b/data/csv/heart-hungarian.csv new file mode 100755 index 0000000..aecefdd --- /dev/null +++ b/data/csv/heart-hungarian.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6555cdc689d4265b8d79bb708f58366a0e13c3ed87d255e0ee2d9d66f290dde +size 32763 diff --git a/data/csv/hepatitis.csv b/data/csv/hepatitis.csv new file mode 100755 index 0000000..66f5365 --- /dev/null +++ b/data/csv/hepatitis.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ac48870fea698f8abb894709027ef1c732eef0df24f6709bb94912d9d3968a8 +size 27231 diff --git a/data/csv/ilpd-indian-liver.csv b/data/csv/ilpd-indian-liver.csv new file mode 100755 index 0000000..5e7c3c0 --- /dev/null +++ b/data/csv/ilpd-indian-liver.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5825ead45a79753bf134becdb0a2e7a482a623137b94a84abaa919825e88b94d +size 50831 diff --git a/data/csv/ionosphere.csv b/data/csv/ionosphere.csv new file mode 100755 index 0000000..2bedc09 --- /dev/null +++ b/data/csv/ionosphere.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb8b1bac4895234172696037c53b1087d648e1ed70d771d557554815767d66fa +size 106782 diff --git a/data/csv/iris.csv b/data/csv/iris.csv new file mode 100755 index 0000000..a58d20c --- /dev/null +++ b/data/csv/iris.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b857ee59af04f3d8a15a1088dc66863871a0d93347cb18a77f344f8e8b86f7d2 +size 5670 diff --git a/data/csv/led-display.csv b/data/csv/led-display.csv new file mode 100755 index 0000000..08d3bdb --- /dev/null +++ b/data/csv/led-display.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77ae16448ccba80ac9a707cb3e1100c600aa59fe1d450c74d4a4a62a5c4358e4 +size 64241 diff --git a/data/csv/libras.csv b/data/csv/libras.csv new file mode 100755 index 0000000..1c38efe --- /dev/null +++ b/data/csv/libras.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd4c3e8a8e52cc3b95c621b212e9d9b6472d731901680e0d04b01e489eaac31e +size 295629 diff --git a/data/csv/low-res-spect.csv b/data/csv/low-res-spect.csv new file mode 100755 index 0000000..284f251 --- /dev/null +++ b/data/csv/low-res-spect.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1be5e667bacd550e415e2d67875a23483b6ef0d0b31fa9b448d460e60e4584ec +size 497350 diff --git a/data/csv/lymphography.csv b/data/csv/lymphography.csv new file mode 100755 index 0000000..0068350 --- /dev/null +++ b/data/csv/lymphography.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d28bcf3da808317bfe996fc318c227f47cca7f9325a11659474919ed917bc0ec +size 24801 diff --git a/data/csv/mammographic.csv b/data/csv/mammographic.csv new file mode 100755 index 0000000..a88db60 --- /dev/null +++ b/data/csv/mammographic.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0935ab0040fb0c5e18ed4ee706d63976e19082d34d971589d52cccd1ac326f3 +size 45398 diff --git a/data/csv/molec-biol-promoter.csv b/data/csv/molec-biol-promoter.csv new file mode 100755 index 0000000..25335b5 --- /dev/null +++ b/data/csv/molec-biol-promoter.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97437ac0be2f5dea923fd9ea56ebbe6526de1eb5431b81c1391867fa36ce9384 +size 53346 diff --git a/data/csv/musk-1.csv b/data/csv/musk-1.csv new file mode 100755 index 0000000..3c78eef --- /dev/null +++ b/data/csv/musk-1.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9aad1f497742988cd654cb60ba89bf8d8459e3ee79273d381bb7365d20d6deda +size 725613 diff --git a/data/csv/oocytes_merluccius_nucleus_4d.csv b/data/csv/oocytes_merluccius_nucleus_4d.csv new file mode 100755 index 0000000..1be5475 --- /dev/null +++ b/data/csv/oocytes_merluccius_nucleus_4d.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b155f5aa776346cbff5d056db1f906d52ffbdf945c10d02e3cebec51123112c3 +size 389155 diff --git a/data/csv/oocytes_merluccius_states_2f.csv b/data/csv/oocytes_merluccius_states_2f.csv new file mode 100755 index 0000000..b18c9a2 --- /dev/null +++ b/data/csv/oocytes_merluccius_states_2f.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc3c2d47acf2d00b22100f388c585c469eeba8174806e63e86c0cf00a565ca19 +size 237384 diff --git a/data/csv/oocytes_trisopterus_nucleus_2f.csv b/data/csv/oocytes_trisopterus_nucleus_2f.csv new file mode 100755 index 0000000..a88b46f --- /dev/null +++ b/data/csv/oocytes_trisopterus_nucleus_2f.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11f3afb414b9936b8bac440e40c5529fb6314a7246f68a8f9bb0b8d6bf14de49 +size 213779 diff --git a/data/csv/oocytes_trisopterus_states_5b.csv b/data/csv/oocytes_trisopterus_states_5b.csv new file mode 100755 index 0000000..7fbf5b3 --- /dev/null +++ b/data/csv/oocytes_trisopterus_states_5b.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2eb2c878860a4de98949e62af0a534b8a31458282939299d4b32fef638944e44 +size 273973 diff --git a/data/csv/parkinsons.csv b/data/csv/parkinsons.csv new file mode 100755 index 0000000..6bbae42 --- /dev/null +++ b/data/csv/parkinsons.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c5816385eb343345b744151a5ddc335b5f9b23d7271352f5485823e15246b16 +size 40601 diff --git a/data/csv/pima.csv b/data/csv/pima.csv new file mode 100755 index 0000000..0bcf273 --- /dev/null +++ b/data/csv/pima.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ab7ca7ac3e0601d2698e0caf6c79913e862fa68596fb41463ddd2689d4aed9c +size 58425 diff --git a/data/csv/pittsburg-bridges-MATERIAL.csv b/data/csv/pittsburg-bridges-MATERIAL.csv new file mode 100755 index 0000000..d0c7c09 --- /dev/null +++ b/data/csv/pittsburg-bridges-MATERIAL.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e41f017c32b190e6aef1173023ae2d7d6e4f5c748f8bbdb2c0ed1eb616d212b +size 6926 diff --git a/data/csv/pittsburg-bridges-REL-L.csv b/data/csv/pittsburg-bridges-REL-L.csv new file mode 100755 index 0000000..35b403d --- /dev/null +++ b/data/csv/pittsburg-bridges-REL-L.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72c156123dab20208f8bce5fd3ab4fb356dc65442e99ed6a86e4a957a5ace303 +size 6680 diff --git a/data/csv/pittsburg-bridges-SPAN.csv b/data/csv/pittsburg-bridges-SPAN.csv new file mode 100755 index 0000000..d87fa54 --- /dev/null +++ b/data/csv/pittsburg-bridges-SPAN.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:551d3a44124aa352223717df3d53db4b0ed662c6e1051c6b898f0427a77d26ad +size 6154 diff --git a/data/csv/pittsburg-bridges-T-OR-D.csv b/data/csv/pittsburg-bridges-T-OR-D.csv new file mode 100755 index 0000000..9ac15db --- /dev/null +++ b/data/csv/pittsburg-bridges-T-OR-D.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:132907f2406759e8bb6391813cfaa1363e95c6b501ef27bc596c4e4479b35f43 +size 6724 diff --git a/data/csv/planning.csv b/data/csv/planning.csv new file mode 100755 index 0000000..8eef770 --- /dev/null +++ b/data/csv/planning.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6754e272d3d13ce785f3bc58189a9bd41d9b8b21a3ef5275f973cf9fe2fce3de +size 20472 diff --git a/data/csv/post-operative.csv b/data/csv/post-operative.csv new file mode 100755 index 0000000..ca4af42 --- /dev/null +++ b/data/csv/post-operative.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24bd14663271fd794dfa925a970e20fe515310baf4e8442d522e5c205b5657eb +size 6820 diff --git a/data/csv/seeds.csv b/data/csv/seeds.csv new file mode 100755 index 0000000..0bc5fad --- /dev/null +++ b/data/csv/seeds.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd95a6d1079b073cf4575b2f40e27b2f384e248406cc94f457f21b87f6de526c +size 13837 diff --git a/data/csv/statlog-australian-credit.csv b/data/csv/statlog-australian-credit.csv new file mode 100755 index 0000000..f51b8b0 --- /dev/null +++ b/data/csv/statlog-australian-credit.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db8b5805c5af83c0ed9155c41df62e962dcc69f9214af8e28870e099cc264103 +size 90041 diff --git a/data/csv/statlog-german-credit.csv b/data/csv/statlog-german-credit.csv new file mode 100755 index 0000000..89cc35c --- /dev/null +++ b/data/csv/statlog-german-credit.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1efb7b844546181ad86b4d2a9e245e2eda6be0b4100b193db82bfc9cc6407f02 +size 220685 diff --git a/data/csv/statlog-heart.csv b/data/csv/statlog-heart.csv new file mode 100755 index 0000000..f0bf5c0 --- /dev/null +++ b/data/csv/statlog-heart.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cee218f7c42475f4339eb2d93ed0adaa105d0213238b7170695a7a83c433a383 +size 32893 diff --git a/data/csv/statlog-image.csv b/data/csv/statlog-image.csv new file mode 100755 index 0000000..dddb99e --- /dev/null +++ b/data/csv/statlog-image.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a4b076c05c9304f06d72d5d77fffa9ebabff0baeb61f8c4517d7d6084ed6499 +size 394674 diff --git a/data/csv/statlog-vehicle.csv b/data/csv/statlog-vehicle.csv new file mode 100755 index 0000000..6c894ef --- /dev/null +++ b/data/csv/statlog-vehicle.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b3b095bfe7a4e7873915a5925bc1983afdde34c4d4c4cefee8e39e911059a3be +size 141365 diff --git a/data/csv/synthetic-control.csv b/data/csv/synthetic-control.csv new file mode 100755 index 0000000..d325d19 --- /dev/null +++ b/data/csv/synthetic-control.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c82d41d84da4a6c919ddcac1ed651c6b744e8fb3fc186cc2eaf05b84f1b55427 +size 328816 diff --git a/data/csv/tic-tac-toe.csv b/data/csv/tic-tac-toe.csv new file mode 100755 index 0000000..f557dfb --- /dev/null +++ b/data/csv/tic-tac-toe.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f65e7bced34139a9b00d4120a4c31bde92727bc3f3373e49eb0a3731dd2e7b5 +size 77514 diff --git a/data/csv/vertebral-column-2clases.csv b/data/csv/vertebral-column-2clases.csv new file mode 100755 index 0000000..8f5c141 --- /dev/null +++ b/data/csv/vertebral-column-2clases.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de94e4e34bbe2d8b2ebd9d4486b3fec80ab364287a8d65752139c053b6e76282 +size 17840 diff --git a/data/csv/wine.csv b/data/csv/wine.csv new file mode 100755 index 0000000..f871410 --- /dev/null +++ b/data/csv/wine.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a39b24ae24b8bd62a7e2f3cf231f07f796b18f6390b012078edb709688e5b05 +size 21564 diff --git a/data/csv/zoo.csv b/data/csv/zoo.csv new file mode 100755 index 0000000..866a521 --- /dev/null +++ b/data/csv/zoo.csv @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:843e4d4b72cc97f08c0af70b6c150ff1ab23e081e5710cc0fd28fa893dc0dae9 +size 15100 diff --git a/data/oc1output/balance-scale.txt b/data/oc1output/balance-scale.txt new file mode 100755 index 0000000..26792de --- /dev/null +++ b/data/oc1output/balance-scale.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 3 +Pruned decision tree written to csv/balance-scale.csv.dt. +Unpruned decision tree written to csv/balance-scale.csv.dt.unpruned. +fold 1: acc. = 91.20 #leaves = 16 max. depth = 9 +fold 2: acc. = 86.40 #leaves = 30 max. depth = 9 +fold 3: acc. = 96.00 #leaves = 20 max. depth = 8 +fold 4: acc. = 92.80 #leaves = 17 max. depth = 5 +fold 5: acc. = 94.40 #leaves = 22 max. depth = 7 +accuracy = 92.16 #leaves = 21.00 max depth = 7.60 +.79400000000000000000 diff --git a/data/oc1output/balloons.txt b/data/oc1output/balloons.txt new file mode 100755 index 0000000..1a9cdaf --- /dev/null +++ b/data/oc1output/balloons.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/balloons.csv.dt. +Unpruned decision tree written to csv/balloons.csv.dt.unpruned. +fold 1: acc. = 66.67 #leaves = 2 max. depth = 1 +fold 2: acc. = 66.67 #leaves = 2 max. depth = 1 +fold 3: acc. = 100.00 #leaves = 4 max. depth = 3 +fold 4: acc. = 100.00 #leaves = 4 max. depth = 3 +fold 5: acc. = 50.00 #leaves = 2 max. depth = 1 +accuracy = 75.00 #leaves = 2.80 max depth = 1.80 +.00400000000000000000 diff --git a/data/oc1output/breast-cancer-wisc-diag.txt b/data/oc1output/breast-cancer-wisc-diag.txt new file mode 100755 index 0000000..93deaf3 --- /dev/null +++ b/data/oc1output/breast-cancer-wisc-diag.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/breast-cancer-wisc-diag.csv.dt. +Unpruned decision tree written to csv/breast-cancer-wisc-diag.csv.dt.unpruned. +fold 1: acc. = 92.92 #leaves = 12 max. depth = 6 +fold 2: acc. = 94.69 #leaves = 11 max. depth = 5 +fold 3: acc. = 92.92 #leaves = 12 max. depth = 5 +fold 4: acc. = 92.04 #leaves = 15 max. depth = 7 +fold 5: acc. = 91.45 #leaves = 7 max. depth = 3 +accuracy = 92.79 #leaves = 11.40 max depth = 5.20 +1.91800000000000000000 diff --git a/data/oc1output/breast-cancer-wisc-prog.txt b/data/oc1output/breast-cancer-wisc-prog.txt new file mode 100755 index 0000000..b6bc2d4 --- /dev/null +++ b/data/oc1output/breast-cancer-wisc-prog.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/breast-cancer-wisc-prog.csv.dt. +Unpruned decision tree written to csv/breast-cancer-wisc-prog.csv.dt.unpruned. +fold 1: acc. = 82.05 #leaves = 2 max. depth = 1 +fold 2: acc. = 71.79 #leaves = 13 max. depth = 10 +fold 3: acc. = 82.05 #leaves = 17 max. depth = 7 +fold 4: acc. = 51.28 #leaves = 16 max. depth = 7 +fold 5: acc. = 78.57 #leaves = 14 max. depth = 8 +accuracy = 73.23 #leaves = 12.40 max depth = 6.60 +.92700000000000000000 diff --git a/data/oc1output/breast-cancer-wisc.txt b/data/oc1output/breast-cancer-wisc.txt new file mode 100755 index 0000000..fa81848 --- /dev/null +++ b/data/oc1output/breast-cancer-wisc.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/breast-cancer-wisc.csv.dt. +Unpruned decision tree written to csv/breast-cancer-wisc.csv.dt.unpruned. +fold 1: acc. = 95.68 #leaves = 4 max. depth = 3 +fold 2: acc. = 93.53 #leaves = 11 max. depth = 7 +fold 3: acc. = 94.24 #leaves = 12 max. depth = 9 +fold 4: acc. = 95.68 #leaves = 15 max. depth = 7 +fold 5: acc. = 94.41 #leaves = 13 max. depth = 8 +accuracy = 94.71 #leaves = 11.00 max depth = 6.80 +1.02400000000000000000 diff --git a/data/oc1output/breast-cancer.txt b/data/oc1output/breast-cancer.txt new file mode 100755 index 0000000..2c862b9 --- /dev/null +++ b/data/oc1output/breast-cancer.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/breast-cancer.csv.dt. +Unpruned decision tree written to csv/breast-cancer.csv.dt.unpruned. +fold 1: acc. = 64.91 #leaves = 4 max. depth = 3 +fold 2: acc. = 56.14 #leaves = 40 max. depth = 12 +fold 3: acc. = 64.91 #leaves = 24 max. depth = 8 +fold 4: acc. = 64.91 #leaves = 37 max. depth = 10 +fold 5: acc. = 70.69 #leaves = 32 max. depth = 10 +accuracy = 64.34 #leaves = 27.40 max depth = 8.60 +.39700000000000000000 diff --git a/data/oc1output/cardiotocography-10clases.txt b/data/oc1output/cardiotocography-10clases.txt new file mode 100755 index 0000000..a895544 --- /dev/null +++ b/data/oc1output/cardiotocography-10clases.txt @@ -0,0 +1,19 @@ +Remapping class numbers: + 8 To 1 + 5 To 2 + 1 To 3 + 7 To 4 + 9 To 5 + 0 To 7 + 2 To 8 + 4 To 9 + 3 To 10 +Pruned decision tree written to csv/cardiotocography-10clases.csv.dt. +Unpruned decision tree written to csv/cardiotocography-10clases.csv.dt.unpruned. +fold 1: acc. = 78.12 #leaves = 19 max. depth = 7 +fold 2: acc. = 78.35 #leaves = 144 max. depth = 15 +fold 3: acc. = 79.29 #leaves = 134 max. depth = 12 +fold 4: acc. = 77.65 #leaves = 138 max. depth = 14 +fold 5: acc. = 80.28 #leaves = 146 max. depth = 14 +accuracy = 78.74 #leaves = 116.20 max depth = 12.40 +20.41800000000000000000 diff --git a/data/oc1output/cardiotocography-3clases.txt b/data/oc1output/cardiotocography-3clases.txt new file mode 100755 index 0000000..5601774 --- /dev/null +++ b/data/oc1output/cardiotocography-3clases.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 2 + 2 To 3 +Pruned decision tree written to csv/cardiotocography-3clases.csv.dt. +Unpruned decision tree written to csv/cardiotocography-3clases.csv.dt.unpruned. +fold 1: acc. = 88.47 #leaves = 14 max. depth = 7 +fold 2: acc. = 91.06 #leaves = 63 max. depth = 12 +fold 3: acc. = 91.76 #leaves = 59 max. depth = 10 +fold 4: acc. = 89.41 #leaves = 58 max. depth = 14 +fold 5: acc. = 90.85 #leaves = 55 max. depth = 12 +accuracy = 90.31 #leaves = 49.80 max depth = 11.00 +11.24900000000000000000 diff --git a/data/oc1output/conn-bench-sonar-mines-rocks.txt b/data/oc1output/conn-bench-sonar-mines-rocks.txt new file mode 100755 index 0000000..b464240 --- /dev/null +++ b/data/oc1output/conn-bench-sonar-mines-rocks.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/conn-bench-sonar-mines-rocks.csv.dt. +Unpruned decision tree written to csv/conn-bench-sonar-mines-rocks.csv.dt.unpruned. +fold 1: acc. = 68.29 #leaves = 2 max. depth = 1 +fold 2: acc. = 70.73 #leaves = 21 max. depth = 9 +fold 3: acc. = 78.05 #leaves = 16 max. depth = 8 +fold 4: acc. = 65.85 #leaves = 12 max. depth = 6 +fold 5: acc. = 65.91 #leaves = 15 max. depth = 6 +accuracy = 69.71 #leaves = 13.20 max depth = 6.00 +.52500000000000000000 diff --git a/data/oc1output/cylinder-bands.txt b/data/oc1output/cylinder-bands.txt new file mode 100755 index 0000000..34f3dc3 --- /dev/null +++ b/data/oc1output/cylinder-bands.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/cylinder-bands.csv.dt. +Unpruned decision tree written to csv/cylinder-bands.csv.dt.unpruned. +fold 1: acc. = 72.55 #leaves = 12 max. depth = 5 +fold 2: acc. = 54.90 #leaves = 41 max. depth = 9 +fold 3: acc. = 65.69 #leaves = 35 max. depth = 10 +fold 4: acc. = 68.63 #leaves = 43 max. depth = 12 +fold 5: acc. = 65.38 #leaves = 35 max. depth = 10 +accuracy = 65.43 #leaves = 33.20 max depth = 9.20 +2.12700000000000000000 diff --git a/data/oc1output/dermatology.txt b/data/oc1output/dermatology.txt new file mode 100755 index 0000000..ac78de9 --- /dev/null +++ b/data/oc1output/dermatology.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 2 + 2 To 3 + 3 To 5 + 5 To 6 +Unpruned decision tree written to csv/dermatology.csv.dt. +fold 1: acc. = 98.63 #leaves = 6 max. depth = 3 +fold 2: acc. = 97.26 #leaves = 6 max. depth = 3 +fold 3: acc. = 93.15 #leaves = 6 max. depth = 3 +fold 4: acc. = 93.15 #leaves = 9 max. depth = 6 +fold 5: acc. = 97.30 #leaves = 8 max. depth = 5 +accuracy = 95.90 #leaves = 7.00 max depth = 4.00 +.54800000000000000000 diff --git a/data/oc1output/echocardiogram.txt b/data/oc1output/echocardiogram.txt new file mode 100755 index 0000000..b35e7cd --- /dev/null +++ b/data/oc1output/echocardiogram.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/echocardiogram.csv.dt. +Unpruned decision tree written to csv/echocardiogram.csv.dt.unpruned. +fold 1: acc. = 84.62 #leaves = 2 max. depth = 1 +fold 2: acc. = 65.38 #leaves = 10 max. depth = 5 +fold 3: acc. = 73.08 #leaves = 8 max. depth = 4 +fold 4: acc. = 61.54 #leaves = 12 max. depth = 7 +fold 5: acc. = 66.67 #leaves = 10 max. depth = 5 +accuracy = 70.23 #leaves = 8.40 max depth = 4.40 +.11500000000000000000 diff --git a/data/oc1output/fertility.txt b/data/oc1output/fertility.txt new file mode 100755 index 0000000..a5ca70f --- /dev/null +++ b/data/oc1output/fertility.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/fertility.csv.dt. +Unpruned decision tree written to csv/fertility.csv.dt.unpruned. +fold 1: acc. = 95.00 #leaves = 2 max. depth = 1 +fold 2: acc. = 70.00 #leaves = 8 max. depth = 4 +fold 3: acc. = 75.00 #leaves = 5 max. depth = 3 +fold 4: acc. = 80.00 #leaves = 9 max. depth = 5 +fold 5: acc. = 75.00 #leaves = 7 max. depth = 4 +accuracy = 79.00 #leaves = 6.20 max depth = 3.40 +.05600000000000000000 diff --git a/data/oc1output/haberman-survival.txt b/data/oc1output/haberman-survival.txt new file mode 100755 index 0000000..a679489 --- /dev/null +++ b/data/oc1output/haberman-survival.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/haberman-survival.csv.dt. +Unpruned decision tree written to csv/haberman-survival.csv.dt.unpruned. +fold 1: acc. = 73.77 #leaves = 2 max. depth = 1 +fold 2: acc. = 60.66 #leaves = 41 max. depth = 13 +fold 3: acc. = 70.49 #leaves = 39 max. depth = 11 +fold 4: acc. = 73.77 #leaves = 39 max. depth = 10 +fold 5: acc. = 61.29 #leaves = 41 max. depth = 9 +accuracy = 67.97 #leaves = 32.40 max depth = 8.80 +.31700000000000000000 diff --git a/data/oc1output/heart-hungarian.txt b/data/oc1output/heart-hungarian.txt new file mode 100755 index 0000000..0790cce --- /dev/null +++ b/data/oc1output/heart-hungarian.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/heart-hungarian.csv.dt. +Unpruned decision tree written to csv/heart-hungarian.csv.dt.unpruned. +fold 1: acc. = 86.21 #leaves = 4 max. depth = 3 +fold 2: acc. = 67.24 #leaves = 19 max. depth = 6 +fold 3: acc. = 68.97 #leaves = 30 max. depth = 9 +fold 4: acc. = 74.14 #leaves = 26 max. depth = 9 +fold 5: acc. = 74.19 #leaves = 19 max. depth = 7 +accuracy = 74.15 #leaves = 19.60 max depth = 6.80 +.48400000000000000000 diff --git a/data/oc1output/hepatitis.txt b/data/oc1output/hepatitis.txt new file mode 100755 index 0000000..da416f0 --- /dev/null +++ b/data/oc1output/hepatitis.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/hepatitis.csv.dt. +Unpruned decision tree written to csv/hepatitis.csv.dt.unpruned. +fold 1: acc. = 74.19 #leaves = 2 max. depth = 1 +fold 2: acc. = 74.19 #leaves = 8 max. depth = 4 +fold 3: acc. = 70.97 #leaves = 11 max. depth = 4 +fold 4: acc. = 80.65 #leaves = 10 max. depth = 4 +fold 5: acc. = 77.42 #leaves = 12 max. depth = 5 +accuracy = 75.48 #leaves = 8.60 max depth = 3.60 +.20400000000000000000 diff --git a/data/oc1output/ilpd-indian-liver.txt b/data/oc1output/ilpd-indian-liver.txt new file mode 100755 index 0000000..f86cad9 --- /dev/null +++ b/data/oc1output/ilpd-indian-liver.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/ilpd-indian-liver.csv.dt. +Unpruned decision tree written to csv/ilpd-indian-liver.csv.dt.unpruned. +fold 1: acc. = 71.55 #leaves = 10 max. depth = 7 +fold 2: acc. = 70.69 #leaves = 49 max. depth = 12 +fold 3: acc. = 68.10 #leaves = 46 max. depth = 12 +fold 4: acc. = 66.38 #leaves = 50 max. depth = 13 +fold 5: acc. = 64.71 #leaves = 53 max. depth = 13 +accuracy = 68.27 #leaves = 41.60 max depth = 11.40 +1.60900000000000000000 diff --git a/data/oc1output/ionosphere.txt b/data/oc1output/ionosphere.txt new file mode 100755 index 0000000..fbac599 --- /dev/null +++ b/data/oc1output/ionosphere.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/ionosphere.csv.dt. +Unpruned decision tree written to csv/ionosphere.csv.dt.unpruned. +fold 1: acc. = 87.14 #leaves = 3 max. depth = 2 +fold 2: acc. = 81.43 #leaves = 17 max. depth = 7 +fold 3: acc. = 84.29 #leaves = 14 max. depth = 7 +fold 4: acc. = 92.86 #leaves = 9 max. depth = 5 +fold 5: acc. = 92.96 #leaves = 12 max. depth = 8 +accuracy = 87.75 #leaves = 11.00 max depth = 5.80 +1.23400000000000000000 diff --git a/data/oc1output/iris.txt b/data/oc1output/iris.txt new file mode 100755 index 0000000..9d0440b --- /dev/null +++ b/data/oc1output/iris.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Unpruned decision tree written to csv/iris.csv.dt. +fold 1: acc. = 90.00 #leaves = 3 max. depth = 2 +fold 2: acc. = 96.67 #leaves = 4 max. depth = 3 +fold 3: acc. = 93.33 #leaves = 5 max. depth = 4 +fold 4: acc. = 96.67 #leaves = 5 max. depth = 4 +fold 5: acc. = 93.33 #leaves = 5 max. depth = 4 +accuracy = 94.00 #leaves = 4.40 max depth = 3.40 +.08200000000000000000 diff --git a/data/oc1output/led-display.txt b/data/oc1output/led-display.txt new file mode 100755 index 0000000..65d6819 --- /dev/null +++ b/data/oc1output/led-display.txt @@ -0,0 +1,19 @@ +Remapping class numbers: + 7 To 1 + 5 To 3 + 0 To 4 + 4 To 5 + 9 To 6 + 8 To 7 + 6 To 8 + 3 To 9 + 1 To 10 +Pruned decision tree written to csv/led-display.csv.dt. +Unpruned decision tree written to csv/led-display.csv.dt.unpruned. +fold 1: acc. = 66.50 #leaves = 14 max. depth = 6 +fold 2: acc. = 74.50 #leaves = 61 max. depth = 9 +fold 3: acc. = 68.00 #leaves = 61 max. depth = 8 +fold 4: acc. = 63.50 #leaves = 62 max. depth = 8 +fold 5: acc. = 67.50 #leaves = 59 max. depth = 8 +accuracy = 68.00 #leaves = 51.40 max depth = 7.80 +1.46800000000000000000 diff --git a/data/oc1output/libras.txt b/data/oc1output/libras.txt new file mode 100755 index 0000000..e4fa510 --- /dev/null +++ b/data/oc1output/libras.txt @@ -0,0 +1,25 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 + 3 To 4 + 4 To 5 + 5 To 6 + 6 To 7 + 7 To 8 + 8 To 9 + 9 To 10 + 10 To 11 + 11 To 12 + 12 To 13 + 13 To 14 + 14 To 15 +Pruned decision tree written to csv/libras.csv.dt. +Unpruned decision tree written to csv/libras.csv.dt.unpruned. +fold 1: acc. = 65.28 #leaves = 22 max. depth = 6 +fold 2: acc. = 58.33 #leaves = 61 max. depth = 9 +fold 3: acc. = 61.11 #leaves = 57 max. depth = 8 +fold 4: acc. = 62.50 #leaves = 61 max. depth = 9 +fold 5: acc. = 62.50 #leaves = 62 max. depth = 10 +accuracy = 61.94 #leaves = 52.60 max depth = 8.40 +2.56300000000000000000 diff --git a/data/oc1output/low-res-spect.txt b/data/oc1output/low-res-spect.txt new file mode 100755 index 0000000..b4c1e19 --- /dev/null +++ b/data/oc1output/low-res-spect.txt @@ -0,0 +1,18 @@ +Remapping class numbers: + 7 To 2 + 8 To 3 + 0 To 4 + 3 To 5 + 2 To 6 + 6 To 7 + 5 To 8 + 4 To 9 +Pruned decision tree written to csv/low-res-spect.csv.dt. +Unpruned decision tree written to csv/low-res-spect.csv.dt.unpruned. +fold 1: acc. = 76.42 #leaves = 9 max. depth = 5 +fold 2: acc. = 85.85 #leaves = 35 max. depth = 8 +fold 3: acc. = 84.91 #leaves = 34 max. depth = 7 +fold 4: acc. = 83.96 #leaves = 35 max. depth = 8 +fold 5: acc. = 83.18 #leaves = 36 max. depth = 7 +accuracy = 82.86 #leaves = 29.80 max depth = 7.00 +5.99000000000000000000 diff --git a/data/oc1output/lymphography.txt b/data/oc1output/lymphography.txt new file mode 100755 index 0000000..91ab876 --- /dev/null +++ b/data/oc1output/lymphography.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 2 To 1 + 1 To 2 + 0 To 4 +Pruned decision tree written to csv/lymphography.csv.dt. +Unpruned decision tree written to csv/lymphography.csv.dt.unpruned. +fold 1: acc. = 75.86 #leaves = 4 max. depth = 3 +fold 2: acc. = 68.97 #leaves = 11 max. depth = 6 +fold 3: acc. = 79.31 #leaves = 10 max. depth = 4 +fold 4: acc. = 75.86 #leaves = 11 max. depth = 5 +fold 5: acc. = 65.62 #leaves = 15 max. depth = 5 +accuracy = 72.97 #leaves = 10.20 max depth = 4.60 +.21600000000000000000 diff --git a/data/oc1output/mammographic.txt b/data/oc1output/mammographic.txt new file mode 100755 index 0000000..5bde662 --- /dev/null +++ b/data/oc1output/mammographic.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/mammographic.csv.dt. +Unpruned decision tree written to csv/mammographic.csv.dt.unpruned. +fold 1: acc. = 83.85 #leaves = 2 max. depth = 1 +fold 2: acc. = 73.44 #leaves = 106 max. depth = 19 +fold 3: acc. = 75.00 #leaves = 104 max. depth = 21 +fold 4: acc. = 78.65 #leaves = 110 max. depth = 13 +fold 5: acc. = 72.54 #leaves = 98 max. depth = 14 +accuracy = 76.69 #leaves = 84.00 max depth = 13.60 +2.02400000000000000000 diff --git a/data/oc1output/molec-biol-promoter.txt b/data/oc1output/molec-biol-promoter.txt new file mode 100755 index 0000000..d231609 --- /dev/null +++ b/data/oc1output/molec-biol-promoter.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/molec-biol-promoter.csv.dt. +Unpruned decision tree written to csv/molec-biol-promoter.csv.dt.unpruned. +fold 1: acc. = 66.67 #leaves = 4 max. depth = 3 +fold 2: acc. = 61.90 #leaves = 13 max. depth = 5 +fold 3: acc. = 71.43 #leaves = 11 max. depth = 5 +fold 4: acc. = 66.67 #leaves = 10 max. depth = 5 +fold 5: acc. = 68.18 #leaves = 11 max. depth = 4 +accuracy = 66.98 #leaves = 9.80 max depth = 4.40 +.01800000000000000000 diff --git a/data/oc1output/musk-1.txt b/data/oc1output/musk-1.txt new file mode 100755 index 0000000..4db1e32 --- /dev/null +++ b/data/oc1output/musk-1.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/musk-1.csv.dt. +Unpruned decision tree written to csv/musk-1.csv.dt.unpruned. +fold 1: acc. = 80.00 #leaves = 14 max. depth = 11 +fold 2: acc. = 83.16 #leaves = 29 max. depth = 9 +fold 3: acc. = 80.00 #leaves = 30 max. depth = 11 +fold 4: acc. = 81.05 #leaves = 34 max. depth = 14 +fold 5: acc. = 80.21 #leaves = 34 max. depth = 16 +accuracy = 80.88 #leaves = 28.20 max depth = 12.20 +3.84800000000000000000 diff --git a/data/oc1output/oocytes_merluccius_nucleus_4d.txt b/data/oc1output/oocytes_merluccius_nucleus_4d.txt new file mode 100755 index 0000000..4489824 --- /dev/null +++ b/data/oc1output/oocytes_merluccius_nucleus_4d.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/oocytes_merluccius_nucleus_4d.csv.dt. +Unpruned decision tree written to csv/oocytes_merluccius_nucleus_4d.csv.dt.unpruned. +fold 1: acc. = 83.82 #leaves = 38 max. depth = 11 +fold 2: acc. = 74.02 #leaves = 81 max. depth = 12 +fold 3: acc. = 71.57 #leaves = 84 max. depth = 13 +fold 4: acc. = 73.53 #leaves = 88 max. depth = 12 +fold 5: acc. = 68.45 #leaves = 85 max. depth = 13 +accuracy = 74.27 #leaves = 75.20 max depth = 12.20 +7.67300000000000000000 diff --git a/data/oc1output/oocytes_merluccius_states_2f.txt b/data/oc1output/oocytes_merluccius_states_2f.txt new file mode 100755 index 0000000..39a21a0 --- /dev/null +++ b/data/oc1output/oocytes_merluccius_states_2f.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 2 To 1 + 1 To 2 + 0 To 3 +Pruned decision tree written to csv/oocytes_merluccius_states_2f.csv.dt. +Unpruned decision tree written to csv/oocytes_merluccius_states_2f.csv.dt.unpruned. +fold 1: acc. = 88.24 #leaves = 5 max. depth = 3 +fold 2: acc. = 87.75 #leaves = 35 max. depth = 9 +fold 3: acc. = 90.69 #leaves = 41 max. depth = 9 +fold 4: acc. = 87.25 #leaves = 35 max. depth = 12 +fold 5: acc. = 85.92 #leaves = 34 max. depth = 11 +accuracy = 87.96 #leaves = 30.00 max depth = 8.80 +5.23000000000000000000 diff --git a/data/oc1output/oocytes_trisopterus_nucleus_2f.txt b/data/oc1output/oocytes_trisopterus_nucleus_2f.txt new file mode 100755 index 0000000..f5f00e1 --- /dev/null +++ b/data/oc1output/oocytes_trisopterus_nucleus_2f.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/oocytes_trisopterus_nucleus_2f.csv.dt. +Unpruned decision tree written to csv/oocytes_trisopterus_nucleus_2f.csv.dt.unpruned. +fold 1: acc. = 75.82 #leaves = 17 max. depth = 8 +fold 2: acc. = 73.63 #leaves = 55 max. depth = 12 +fold 3: acc. = 70.33 #leaves = 70 max. depth = 15 +fold 4: acc. = 70.33 #leaves = 59 max. depth = 10 +fold 5: acc. = 74.46 #leaves = 70 max. depth = 11 +accuracy = 72.92 #leaves = 54.20 max depth = 11.20 +4.95800000000000000000 diff --git a/data/oc1output/oocytes_trisopterus_states_5b.txt b/data/oc1output/oocytes_trisopterus_states_5b.txt new file mode 100755 index 0000000..ba0b407 --- /dev/null +++ b/data/oc1output/oocytes_trisopterus_states_5b.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 2 + 2 To 3 +Pruned decision tree written to csv/oocytes_trisopterus_states_5b.csv.dt. +Unpruned decision tree written to csv/oocytes_trisopterus_states_5b.csv.dt.unpruned. +fold 1: acc. = 86.81 #leaves = 7 max. depth = 4 +fold 2: acc. = 84.62 #leaves = 45 max. depth = 19 +fold 3: acc. = 84.07 #leaves = 41 max. depth = 12 +fold 4: acc. = 93.96 #leaves = 48 max. depth = 16 +fold 5: acc. = 81.52 #leaves = 42 max. depth = 12 +accuracy = 86.18 #leaves = 36.60 max depth = 12.60 +7.58400000000000000000 diff --git a/data/oc1output/parkinsons.txt b/data/oc1output/parkinsons.txt new file mode 100755 index 0000000..d66f8f9 --- /dev/null +++ b/data/oc1output/parkinsons.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/parkinsons.csv.dt. +Unpruned decision tree written to csv/parkinsons.csv.dt.unpruned. +fold 1: acc. = 89.74 #leaves = 4 max. depth = 3 +fold 2: acc. = 92.31 #leaves = 7 max. depth = 4 +fold 3: acc. = 84.62 #leaves = 8 max. depth = 6 +fold 4: acc. = 76.92 #leaves = 8 max. depth = 6 +fold 5: acc. = 89.74 #leaves = 7 max. depth = 5 +accuracy = 86.67 #leaves = 6.80 max depth = 4.80 +.47200000000000000000 diff --git a/data/oc1output/pima.txt b/data/oc1output/pima.txt new file mode 100755 index 0000000..710db5d --- /dev/null +++ b/data/oc1output/pima.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/pima.csv.dt. +Unpruned decision tree written to csv/pima.csv.dt.unpruned. +fold 1: acc. = 71.90 #leaves = 6 max. depth = 5 +fold 2: acc. = 70.59 #leaves = 70 max. depth = 11 +fold 3: acc. = 70.59 #leaves = 61 max. depth = 12 +fold 4: acc. = 71.24 #leaves = 67 max. depth = 11 +fold 5: acc. = 66.67 #leaves = 60 max. depth = 13 +accuracy = 70.18 #leaves = 52.80 max depth = 10.40 +1.88100000000000000000 diff --git a/data/oc1output/pittsburg-bridges-MATERIAL.txt b/data/oc1output/pittsburg-bridges-MATERIAL.txt new file mode 100755 index 0000000..e9609b7 --- /dev/null +++ b/data/oc1output/pittsburg-bridges-MATERIAL.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Pruned decision tree written to csv/pittsburg-bridges-MATERIAL.csv.dt. +Unpruned decision tree written to csv/pittsburg-bridges-MATERIAL.csv.dt.unpruned. +fold 1: acc. = 71.43 #leaves = 3 max. depth = 2 +fold 2: acc. = 61.90 #leaves = 8 max. depth = 5 +fold 3: acc. = 80.95 #leaves = 8 max. depth = 4 +fold 4: acc. = 85.71 #leaves = 12 max. depth = 6 +fold 5: acc. = 95.45 #leaves = 9 max. depth = 5 +accuracy = 79.25 #leaves = 8.00 max depth = 4.40 +.07500000000000000000 diff --git a/data/oc1output/pittsburg-bridges-REL-L.txt b/data/oc1output/pittsburg-bridges-REL-L.txt new file mode 100755 index 0000000..f767a6d --- /dev/null +++ b/data/oc1output/pittsburg-bridges-REL-L.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Pruned decision tree written to csv/pittsburg-bridges-REL-L.csv.dt. +Unpruned decision tree written to csv/pittsburg-bridges-REL-L.csv.dt.unpruned. +fold 1: acc. = 70.00 #leaves = 3 max. depth = 2 +fold 2: acc. = 70.00 #leaves = 17 max. depth = 8 +fold 3: acc. = 70.00 #leaves = 16 max. depth = 7 +fold 4: acc. = 70.00 #leaves = 12 max. depth = 4 +fold 5: acc. = 60.87 #leaves = 14 max. depth = 7 +accuracy = 67.96 #leaves = 12.40 max depth = 5.60 +.12200000000000000000 diff --git a/data/oc1output/pittsburg-bridges-SPAN.txt b/data/oc1output/pittsburg-bridges-SPAN.txt new file mode 100755 index 0000000..cbd5997 --- /dev/null +++ b/data/oc1output/pittsburg-bridges-SPAN.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Pruned decision tree written to csv/pittsburg-bridges-SPAN.csv.dt. +Unpruned decision tree written to csv/pittsburg-bridges-SPAN.csv.dt.unpruned. +fold 1: acc. = 55.56 #leaves = 2 max. depth = 1 +fold 2: acc. = 72.22 #leaves = 19 max. depth = 8 +fold 3: acc. = 61.11 #leaves = 15 max. depth = 7 +fold 4: acc. = 50.00 #leaves = 9 max. depth = 5 +fold 5: acc. = 70.00 #leaves = 11 max. depth = 6 +accuracy = 61.96 #leaves = 11.20 max depth = 5.40 +.10200000000000000000 diff --git a/data/oc1output/pittsburg-bridges-T-OR-D.txt b/data/oc1output/pittsburg-bridges-T-OR-D.txt new file mode 100755 index 0000000..6e73a13 --- /dev/null +++ b/data/oc1output/pittsburg-bridges-T-OR-D.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/pittsburg-bridges-T-OR-D.csv.dt. +Unpruned decision tree written to csv/pittsburg-bridges-T-OR-D.csv.dt.unpruned. +fold 1: acc. = 65.00 #leaves = 2 max. depth = 1 +fold 2: acc. = 75.00 #leaves = 9 max. depth = 6 +fold 3: acc. = 85.00 #leaves = 8 max. depth = 5 +fold 4: acc. = 85.00 #leaves = 9 max. depth = 6 +fold 5: acc. = 81.82 #leaves = 9 max. depth = 5 +accuracy = 78.43 #leaves = 7.40 max depth = 4.60 +.08400000000000000000 diff --git a/data/oc1output/planning.txt b/data/oc1output/planning.txt new file mode 100755 index 0000000..1a0cc53 --- /dev/null +++ b/data/oc1output/planning.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/planning.csv.dt. +Unpruned decision tree written to csv/planning.csv.dt.unpruned. +fold 1: acc. = 75.00 #leaves = 2 max. depth = 1 +fold 2: acc. = 61.11 #leaves = 20 max. depth = 9 +fold 3: acc. = 58.33 #leaves = 23 max. depth = 8 +fold 4: acc. = 63.89 #leaves = 22 max. depth = 7 +fold 5: acc. = 55.26 #leaves = 22 max. depth = 8 +accuracy = 62.64 #leaves = 17.80 max depth = 6.60 +.35900000000000000000 diff --git a/data/oc1output/post-operative.txt b/data/oc1output/post-operative.txt new file mode 100755 index 0000000..3258a26 --- /dev/null +++ b/data/oc1output/post-operative.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 3 +Pruned decision tree written to csv/post-operative.csv.dt. +Unpruned decision tree written to csv/post-operative.csv.dt.unpruned. +fold 1: acc. = 72.22 #leaves = 8 max. depth = 4 +fold 2: acc. = 50.00 #leaves = 20 max. depth = 7 +fold 3: acc. = 55.56 #leaves = 17 max. depth = 9 +fold 4: acc. = 55.56 #leaves = 13 max. depth = 5 +fold 5: acc. = 61.11 #leaves = 16 max. depth = 8 +accuracy = 58.89 #leaves = 14.80 max depth = 6.60 +.09100000000000000000 diff --git a/data/oc1output/seeds.txt b/data/oc1output/seeds.txt new file mode 100755 index 0000000..33be9d3 --- /dev/null +++ b/data/oc1output/seeds.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Pruned decision tree written to csv/seeds.csv.dt. +Unpruned decision tree written to csv/seeds.csv.dt.unpruned. +fold 1: acc. = 95.24 #leaves = 3 max. depth = 2 +fold 2: acc. = 95.24 #leaves = 6 max. depth = 5 +fold 3: acc. = 92.86 #leaves = 4 max. depth = 3 +fold 4: acc. = 92.86 #leaves = 7 max. depth = 5 +fold 5: acc. = 92.86 #leaves = 6 max. depth = 3 +accuracy = 93.81 #leaves = 5.20 max depth = 3.60 +.17400000000000000000 diff --git a/data/oc1output/statlog-australian-credit.txt b/data/oc1output/statlog-australian-credit.txt new file mode 100755 index 0000000..c8a463e --- /dev/null +++ b/data/oc1output/statlog-australian-credit.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/statlog-australian-credit.csv.dt. +Unpruned decision tree written to csv/statlog-australian-credit.csv.dt.unpruned. +fold 1: acc. = 60.14 #leaves = 3 max. depth = 2 +fold 2: acc. = 59.42 #leaves = 64 max. depth = 12 +fold 3: acc. = 57.97 #leaves = 61 max. depth = 12 +fold 4: acc. = 59.42 #leaves = 66 max. depth = 13 +fold 5: acc. = 59.42 #leaves = 58 max. depth = 13 +accuracy = 59.28 #leaves = 50.40 max depth = 10.40 +2.20300000000000000000 diff --git a/data/oc1output/statlog-german-credit.txt b/data/oc1output/statlog-german-credit.txt new file mode 100755 index 0000000..d55ac37 --- /dev/null +++ b/data/oc1output/statlog-german-credit.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/statlog-german-credit.csv.dt. +Unpruned decision tree written to csv/statlog-german-credit.csv.dt.unpruned. +fold 1: acc. = 70.50 #leaves = 2 max. depth = 1 +fold 2: acc. = 65.00 #leaves = 81 max. depth = 14 +fold 3: acc. = 71.50 #leaves = 69 max. depth = 12 +fold 4: acc. = 67.00 #leaves = 67 max. depth = 12 +fold 5: acc. = 61.50 #leaves = 64 max. depth = 14 +accuracy = 67.10 #leaves = 56.60 max depth = 10.60 +4.37200000000000000000 diff --git a/data/oc1output/statlog-heart.txt b/data/oc1output/statlog-heart.txt new file mode 100755 index 0000000..7355fd6 --- /dev/null +++ b/data/oc1output/statlog-heart.txt @@ -0,0 +1,10 @@ +Remapping class numbers: + 0 To 2 +Unpruned decision tree written to csv/statlog-heart.csv.dt. +fold 1: acc. = 74.07 #leaves = 13 max. depth = 7 +fold 2: acc. = 81.48 #leaves = 15 max. depth = 7 +fold 3: acc. = 68.52 #leaves = 15 max. depth = 9 +fold 4: acc. = 74.07 #leaves = 16 max. depth = 6 +fold 5: acc. = 61.11 #leaves = 14 max. depth = 6 +accuracy = 71.85 #leaves = 14.60 max depth = 7.00 +.43200000000000000000 diff --git a/data/oc1output/statlog-image.txt b/data/oc1output/statlog-image.txt new file mode 100755 index 0000000..066f69a --- /dev/null +++ b/data/oc1output/statlog-image.txt @@ -0,0 +1,16 @@ +Remapping class numbers: + 5 To 1 + 1 To 3 + 6 To 4 + 0 To 5 + 3 To 6 + 4 To 7 +Pruned decision tree written to csv/statlog-image.csv.dt. +Unpruned decision tree written to csv/statlog-image.csv.dt.unpruned. +fold 1: acc. = 94.16 #leaves = 38 max. depth = 8 +fold 2: acc. = 96.32 #leaves = 40 max. depth = 12 +fold 3: acc. = 92.86 #leaves = 49 max. depth = 10 +fold 4: acc. = 96.54 #leaves = 44 max. depth = 11 +fold 5: acc. = 92.21 #leaves = 44 max. depth = 9 +accuracy = 94.42 #leaves = 43.00 max depth = 10.00 +11.76100000000000000000 diff --git a/data/oc1output/statlog-vehicle.txt b/data/oc1output/statlog-vehicle.txt new file mode 100755 index 0000000..5ec26a1 --- /dev/null +++ b/data/oc1output/statlog-vehicle.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 3 To 1 + 0 To 3 + 1 To 4 +Unpruned decision tree written to csv/statlog-vehicle.csv.dt. +fold 1: acc. = 71.01 #leaves = 75 max. depth = 13 +fold 2: acc. = 77.51 #leaves = 81 max. depth = 15 +fold 3: acc. = 66.86 #leaves = 76 max. depth = 12 +fold 4: acc. = 74.56 #leaves = 76 max. depth = 13 +fold 5: acc. = 68.82 #leaves = 65 max. depth = 13 +accuracy = 71.75 #leaves = 74.60 max depth = 13.20 +4.58300000000000000000 diff --git a/data/oc1output/synthetic-control.txt b/data/oc1output/synthetic-control.txt new file mode 100755 index 0000000..e029ad2 --- /dev/null +++ b/data/oc1output/synthetic-control.txt @@ -0,0 +1,16 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 + 3 To 4 + 4 To 5 + 5 To 6 +Pruned decision tree written to csv/synthetic-control.csv.dt. +Unpruned decision tree written to csv/synthetic-control.csv.dt.unpruned. +fold 1: acc. = 82.50 #leaves = 11 max. depth = 6 +fold 2: acc. = 85.83 #leaves = 21 max. depth = 7 +fold 3: acc. = 83.33 #leaves = 21 max. depth = 6 +fold 4: acc. = 82.50 #leaves = 24 max. depth = 6 +fold 5: acc. = 81.67 #leaves = 25 max. depth = 7 +accuracy = 83.17 #leaves = 20.40 max depth = 6.40 +7.25500000000000000000 diff --git a/data/oc1output/tic-tac-toe.txt b/data/oc1output/tic-tac-toe.txt new file mode 100755 index 0000000..a872c60 --- /dev/null +++ b/data/oc1output/tic-tac-toe.txt @@ -0,0 +1,11 @@ +Remapping class numbers: + 0 To 2 +Pruned decision tree written to csv/tic-tac-toe.csv.dt. +Unpruned decision tree written to csv/tic-tac-toe.csv.dt.unpruned. +fold 1: acc. = 91.62 #leaves = 11 max. depth = 5 +fold 2: acc. = 91.62 #leaves = 21 max. depth = 8 +fold 3: acc. = 89.01 #leaves = 28 max. depth = 11 +fold 4: acc. = 94.24 #leaves = 25 max. depth = 9 +fold 5: acc. = 94.33 #leaves = 21 max. depth = 8 +accuracy = 92.17 #leaves = 21.20 max depth = 8.20 +1.75200000000000000000 diff --git a/data/oc1output/vertebral-column-2clases.txt b/data/oc1output/vertebral-column-2clases.txt new file mode 100755 index 0000000..8428aaa --- /dev/null +++ b/data/oc1output/vertebral-column-2clases.txt @@ -0,0 +1,12 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 +Pruned decision tree written to csv/vertebral-column-2clases.csv.dt. +Unpruned decision tree written to csv/vertebral-column-2clases.csv.dt.unpruned. +fold 1: acc. = 87.10 #leaves = 5 max. depth = 4 +fold 2: acc. = 72.58 #leaves = 19 max. depth = 7 +fold 3: acc. = 79.03 #leaves = 21 max. depth = 8 +fold 4: acc. = 80.65 #leaves = 18 max. depth = 7 +fold 5: acc. = 82.26 #leaves = 19 max. depth = 7 +accuracy = 80.32 #leaves = 16.40 max depth = 6.60 +.38200000000000000000 diff --git a/data/oc1output/wine.txt b/data/oc1output/wine.txt new file mode 100755 index 0000000..6409aaa --- /dev/null +++ b/data/oc1output/wine.txt @@ -0,0 +1,13 @@ +Remapping class numbers: + 0 To 1 + 1 To 2 + 2 To 3 +Pruned decision tree written to csv/wine.csv.dt. +Unpruned decision tree written to csv/wine.csv.dt.unpruned. +fold 1: acc. = 85.71 #leaves = 3 max. depth = 2 +fold 2: acc. = 94.29 #leaves = 4 max. depth = 3 +fold 3: acc. = 97.14 #leaves = 4 max. depth = 3 +fold 4: acc. = 94.29 #leaves = 4 max. depth = 3 +fold 5: acc. = 92.11 #leaves = 4 max. depth = 3 +accuracy = 92.70 #leaves = 3.80 max depth = 2.80 +.17000000000000000000 diff --git a/data/oc1output/zoo.txt b/data/oc1output/zoo.txt new file mode 100755 index 0000000..78d3a7f --- /dev/null +++ b/data/oc1output/zoo.txt @@ -0,0 +1,16 @@ +Remapping class numbers: + 0 To 1 + 3 To 2 + 1 To 3 + 6 To 4 + 4 To 6 + 2 To 7 +Pruned decision tree written to csv/zoo.csv.dt. +Unpruned decision tree written to csv/zoo.csv.dt.unpruned. +fold 1: acc. = 85.00 #leaves = 5 max. depth = 4 +fold 2: acc. = 80.00 #leaves = 9 max. depth = 6 +fold 3: acc. = 85.00 #leaves = 7 max. depth = 4 +fold 4: acc. = 90.00 #leaves = 10 max. depth = 7 +fold 5: acc. = 85.71 #leaves = 10 max. depth = 7 +accuracy = 85.15 #leaves = 8.20 max depth = 5.60 +.07500000000000000000 diff --git a/experimentation/Database.py b/experimentation/Database.py index 505b53b..b4014ee 100644 --- a/experimentation/Database.py +++ b/experimentation/Database.py @@ -44,7 +44,7 @@ class MySQL: self._database = mysql.connector.connect(**self._config_db) return self._database - def find_best(self, dataset, classifier="any"): + def find_best(self, dataset, classifier="any", experiment="any"): cursor = self._database.cursor(buffered=True) if classifier == "any": command = ( @@ -58,6 +58,7 @@ class MySQL: f"r.dataset=e.dataset where r.dataset='{dataset}' and " f"classifier='{classifier}' and date>='2021-01-20'" ) + command += "" if experiment == "any" else f" and type='{experiment}'" command += ( " order by r.dataset, accuracy desc, classifier desc, " "type, date, time" @@ -163,7 +164,14 @@ class BD(ABC): self.execute(command) def mirror( - self, exp_type, dataset, normalize, standardize, accuracy, parameters + self, + exp_type, + dataset, + normalize, + standardize, + accuracy, + time_spent, + parameters, ) -> None: """Create a record in MySQL database @@ -174,8 +182,8 @@ class BD(ABC): database = dbh.get_connection() command_insert = ( "replace into results (date, time, type, accuracy, " - "dataset, classifier, norm, stand, parameters) values (%s, %s, " - "%s, %s, %s, %s, %s, %s, %s)" + "dataset, classifier, norm, stand, parameters, accuracy_std, time_spent, time_spent_std) values (%s, %s, " + "%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)" ) now = datetime.now() date = now.strftime("%Y-%m-%d") @@ -184,12 +192,15 @@ class BD(ABC): date, time, exp_type, - accuracy, + accuracy[0], dataset, self._model, normalize, standardize, parameters, + accuracy[1], + time_spent[0], + time_spent[1], ) cursor = database.cursor() cursor.execute(command_insert, values) @@ -325,7 +336,14 @@ class Outcomes(BD): dataset, normalize, standardize, - float(results["test_score"].mean()), + [ + float(results["test_score"].mean()), + float(results["test_score"].std()), + ], + [ + float(results["score_time"].mean()), + float(results["score_time"].std()), + ], parameters, ) @@ -414,9 +432,22 @@ class Hyperparameters(BD): command += ") " command_values += ")" self.execute(command + command_values) - accuracy = float(outcomes["test_score"]) + accuracy = [ + float(outcomes["test_score"]), + float(outcomes["test_score_std"]), + ] + time_spent = [ + float(outcomes["score_time"]), + float(outcomes["score_time_std"]), + ] self.mirror( - grid_type, dataset, normalize, standardize, accuracy, parameters + grid_type, + dataset, + normalize, + standardize, + accuracy, + time_spent, + parameters, ) def report(self, dataset, exclude_params): diff --git a/oc1datasets.py b/oc1datasets.py new file mode 100755 index 0000000..828ed40 --- /dev/null +++ b/oc1datasets.py @@ -0,0 +1,13 @@ +import os +import pandas as pd +from experimentation.Sets import Datasets + +dt = Datasets(normalize=True, set_of_files="tanveer") +for data in dt: + name = data[0] + X, y = dt.load(name) + dataset = pd.DataFrame(X) + dataset["y"] = y + file_name = os.path.join("data", "csv", f"{name}.csv") + print(f"{file_name}") + pd.DataFrame(dataset).to_csv(file_name, index=False, sep=",", header=False) diff --git a/oc1results.py b/oc1results.py new file mode 100755 index 0000000..4be1dfc --- /dev/null +++ b/oc1results.py @@ -0,0 +1,51 @@ +import os +from datetime import datetime +from experimentation.Sets import Datasets +from experimentation.Database import MySQL + + +def store_result(database, dataset, accuracy, time_spent): + command_insert = ( + "replace into results (date, time, type, accuracy, " + "dataset, classifier, norm, stand, time_spent, parameters) values (%s," + " %s, %s, %s, %s, %s, %s, %s, %s, %s)" + ) + now = datetime.now() + date = now.strftime("%Y-%m-%d") + time = now.strftime("%H:%M:%S") + values = ( + date, + time, + "crossval", + accuracy, + dataset, + "oc1", + True, + False, + time_spent, + str({"random_state": 1}), + ) + cursor = database.cursor() + cursor.execute(command_insert, values) + database.commit() + + +def get_result(name): + filename = os.path.join("data", "oc1output", f"{name}.txt") + result_file = open(filename, "r") + lines = result_file.readlines() + accuracy = float(lines[-2].split("= ")[1].split("#")[0]) / 100 + time = float(lines[-1]) + print(f"accuracy=[{accuracy}] time=[{time}]") + result_file.close() + return accuracy, time + + +dbh = MySQL() +database = dbh.get_connection() +dt = Datasets(False, False, "tanveer") +for dataset in dt: + print(f"Processing {dataset[0]:30s}", end=" ") + accuracy, time_spent = get_result(dataset[0]) + store_result(database, dataset[0], accuracy, time_spent) +dbh.close() diff --git a/report_mysql.py b/report_mysql.py index cba707f..384dcbd 100644 --- a/report_mysql.py +++ b/report_mysql.py @@ -4,7 +4,7 @@ from experimentation.Sets import Datasets from experimentation.Utils import TextColor from experimentation.Database import MySQL -models = ["stree", "adaBoost", "bagging", "odte"] +models = ["stree", "adaBoost", "bagging", "odte", "cart", "oc1"] def parse_arguments() -> Tuple[str, str, str, bool, bool]: diff --git a/requirements.txt b/requirements.txt index 96aafed..ad28742 100644 --- a/requirements.txt +++ b/requirements.txt @@ -2,8 +2,6 @@ numpy scikit-learn pandas ipympl -myql-client +mysql-client mysql-connector sshtunnel -git+https://github.com/doctorado-ml/stree -git+https://github.com/doctorado-ml/odte diff --git a/results/adaBoost.txt b/results/adaBoost.txt new file mode 100755 index 0000000..6f9e9ba --- /dev/null +++ b/results/adaBoost.txt @@ -0,0 +1,62 @@ +****************************************************************************************************** +* Best Hyperparameters found for datasets -- adaBoost classifier -- * +****************************************************************************************************** + +Date Time Type Classifier Dataset Nor Std Accuracy Reference +========== ======== ========== ========== ============================== === === ========= =========== +2020-11-16 13:22:39 gridsearch adaBoost balance-scale 1 1 0.9344000 0.9046280 + +2020-11-21 23:48:05 gridsearch adaBoost balloons 1 0 0.8833330 0.6625000 + +2020-12-10 20:44:44 gridsearch adaBoost breast-cancer-wisc-diag 1 0 0.9754390 0.9743450 + +2020-11-21 23:51:46 gridsearch adaBoost breast-cancer-wisc-prog 1 0 0.8283330 0.7993400 + +2020-11-21 23:53:27 gridsearch adaBoost breast-cancer-wisc 1 0 0.9685610 0.9702560 - +2020-11-21 23:55:47 gridsearch adaBoost breast-cancer 1 0 0.7273440 0.7382400 - +2020-11-22 2:21:32 gridsearch adaBoost cardiotocography-10clases 1 0 0.6848680 0.8277610 - +2020-12-11 9:28:08 gridsearch adaBoost cardiotocography-3clases 1 0 0.8504380 0.9201340 - +2020-12-10 20:24:21 gridsearch adaBoost conn-bench-sonar-mines-rocks 1 0 0.6876890 0.8336540 - +2020-12-10 22:08:18 gridsearch adaBoost cylinder-bands 1 0 0.6504280 0.7691410 - +2020-11-21 23:52:12 gridsearch adaBoost dermatology 1 0 0.9727140 0.9732780 - +2020-11-21 23:48:42 gridsearch adaBoost echocardiogram 1 0 0.8544160 0.8485270 + +2020-11-16 13:22:40 gridsearch adaBoost fertility 1 1 0.8800000 0.8840000 - +2020-11-22 0:07:17 gridsearch adaBoost haberman-survival 1 0 0.7581700 0.7392540 + +2020-11-21 23:54:11 gridsearch adaBoost heart-hungarian 1 0 0.8400930 0.8204750 + +2020-11-21 23:48:31 gridsearch adaBoost hepatitis 1 0 0.8774190 0.8232030 + +2020-11-21 23:57:45 gridsearch adaBoost ilpd-indian-liver 1 0 0.7376070 0.7150280 + +2020-12-10 20:25:26 gridsearch adaBoost ionosphere 1 0 0.9430180 0.9442150 - +2020-06-26 18:09:17 gridsearch adaBoost iris 1 0 0.9800000 0.9786560 + +2020-11-21 23:54:15 gridsearch adaBoost led-display 1 0 0.7150000 0.7102000 + +2020-12-10 21:21:15 gridsearch adaBoost libras 1 0 0.7277780 0.8911110 - +2020-12-10 20:28:48 gridsearch adaBoost low-res-spect 1 0 0.8832660 0.9028200 - +2020-11-21 23:49:21 gridsearch adaBoost lymphography 1 0 0.8643680 0.8554050 + +2020-11-22 1:01:46 gridsearch adaBoost mammographic 1 0 0.8283090 0.8274720 + +2020-12-10 20:10:27 gridsearch adaBoost molec-biol-promoter 1 0 0.8112550 0.8182690 - +2020-12-10 22:30:56 gridsearch adaBoost musk-1 1 0 0.8256360 0.8764710 - +2020-11-22 12:34:38 gridsearch adaBoost oocytes_merluccius_nucleus_4d 1 0 0.8346200 0.8399630 - +2020-11-22 0:11:40 gridsearch adaBoost oocytes_merluccius_states_2f 1 0 0.9236490 0.9299630 - +2020-11-22 0:49:21 gridsearch adaBoost oocytes_trisopterus_nucleus_2f 1 0 0.7401250 0.8333330 - +2020-11-22 0:22:51 gridsearch adaBoost oocytes_trisopterus_states_5b 1 0 0.8541340 0.9315790 - +2020-11-21 23:53:57 gridsearch adaBoost parkinsons 1 0 0.8564100 0.9202210 - +2020-11-22 0:24:41 gridsearch adaBoost pima 1 0 0.7747390 0.7671880 + +2020-11-21 23:55:43 gridsearch adaBoost pittsburg-bridges-MATERIAL 1 0 0.8675320 0.8642860 + +2020-11-21 23:55:40 gridsearch adaBoost pittsburg-bridges-REL-L 1 0 0.6976190 0.6959290 + +2020-11-21 23:56:48 gridsearch adaBoost pittsburg-bridges-SPAN 1 0 0.7321640 0.6891300 + +2020-11-22 0:00:29 gridsearch adaBoost pittsburg-bridges-T-OR-D 1 0 0.8919050 0.8743700 + +2020-11-21 23:58:28 gridsearch adaBoost planning 1 0 0.7256760 0.7255790 + +2020-11-21 23:57:36 gridsearch adaBoost post-operative 1 0 0.7222220 0.7117420 + +2020-11-21 23:57:45 gridsearch adaBoost seeds 1 0 0.9428570 0.9563030 - +2020-11-22 0:26:38 gridsearch adaBoost statlog-australian-credit 1 0 0.6869570 0.6782810 + +2020-11-22 0:35:42 gridsearch adaBoost statlog-german-credit 1 0 0.7690000 0.7562000 + +2020-11-22 0:03:22 gridsearch adaBoost statlog-heart 1 0 0.8407410 0.8422990 - +2020-11-22 0:29:34 gridsearch adaBoost statlog-image 1 0 0.9593070 0.9761940 - +2020-11-22 1:03:33 gridsearch adaBoost statlog-vehicle 1 0 0.7860700 0.8006730 - +2020-12-10 20:32:31 gridsearch adaBoost synthetic-control 1 0 0.9816670 0.9903330 - +2020-11-22 0:29:36 gridsearch adaBoost tic-tac-toe 1 0 0.9853400 0.9853850 - +2020-11-22 0:13:09 gridsearch adaBoost vertebral-column-2clases 1 0 0.8387100 0.8491530 - +2020-11-22 0:08:20 gridsearch adaBoost wine 1 0 0.9888890 0.9932810 - +2020-11-22 0:08:46 gridsearch adaBoost zoo 1 0 0.9600000 0.9603850 - +we have better results 22 times +we have worse results 27 times +we have equal results 0 times +stree used 0 times +bagging used 0 times +adaBoost used 49 times +odte used 0 times \ No newline at end of file diff --git a/results/analysis.txt b/results/analysis.txt new file mode 100755 index 0000000..f1df93b --- /dev/null +++ b/results/analysis.txt @@ -0,0 +1,61 @@ +**************************************************************************************** +* Best model results * +**************************************************************************************** + +Dataset Reference stree odte adaBoost bagging +============================== ========= =========== =========== =========== =========== +balance-scale 0.904628 0.9488 + 0.9712 + 0.9344 + 0.8352 - +balloons 0.6625 0.866667 + 0.866667 + 0.883333 + 0.683333 + +breast-cancer-wisc-diag 0.974345 0.978932 + 0.982456 + 0.975439 + 0.980702 + +breast-cancer-wisc-prog 0.79934 0.828462 + 0.848718 + 0.828333 + 0.833718 + +breast-cancer-wisc 0.970256 0.965694 - 0.974265 + 0.968561 - 0.977122 + +breast-cancer 0.73824 0.730853 - 0.744949 + 0.727344 - 0.751906 + +cardiotocography-10clases 0.827761 0.666522 - 0.751651 - 0.684868 - 0.702258 - +cardiotocography-3clases 0.920134 0.848074 - 0.861255 - 0.850438 - 0.849022 - +conn-bench-sonar-mines-rocks 0.833654 0.688269 - 0.755052 - 0.687689 - 0.73043 - +cylinder-bands 0.769141 0.632667 - 0.66996 - 0.650428 - 0.666134 - +dermatology 0.973278 0.975454 + 0.986412 + 0.972714 - 0.978156 + +echocardiogram 0.848527 0.847293 - 0.869801 + 0.854416 + 0.869801 + +fertility 0.884 0.88 - 0.89 + 0.88 - 0.89 + +haberman-survival 0.739254 0.764675 + 0.774564 + 0.75817 + 0.738604 - +heart-hungarian 0.820475 0.829924 + 0.843717 + 0.840093 + 0.847107 + +hepatitis 0.823203 0.864516 + 0.877419 + 0.877419 + 0.864516 + +ilpd-indian-liver 0.715028 0.742691 + 0.737577 + 0.737607 + 0.730769 + +ionosphere 0.944215 0.948732 + 0.957223 + 0.943018 - 0.960081 + +iris 0.978656 0.98 + 0.993333 + 0.98 + 0.98 + +led-display 0.7102 0.712 + 0.731 + 0.715 + 0.695 - +libras 0.891111 0.702778 - 0.802778 - 0.727778 - 0.819444 - +low-res-spect 0.90282 0.879492 - 0.913402 + 0.883266 - 0.913384 + +lymphography 0.855405 0.864828 + 0.891724 + 0.864368 + 0.891954 + +mammographic 0.827472 0.829372 + 0.842876 + 0.828309 + 0.837689 + +molec-biol-promoter 0.818269 0.810822 - 0.896104 + 0.811255 - 0.942857 + +musk-1 0.876471 0.789912 - 0.833903 - 0.825636 - 0.817039 - +oocytes_merluccius_nucleus_4d 0.839963 0.808221 - 0.843419 + 0.83462 - 0.792573 - +oocytes_merluccius_states_2f 0.929963 0.911903 - 0.929512 - 0.923649 - 0.729928 - +oocytes_trisopterus_nucleus_2f 0.833333 0.747691 - 0.750964 - 0.740125 - 0.727971 - +oocytes_trisopterus_states_5b 0.931579 0.845361 - 0.876136 - 0.854134 - 0.85408 - +parkinsons 0.920221 0.846154 - 0.866667 - 0.85641 - 0.861538 - +pima 0.767188 0.780002 + 0.789093 + 0.774739 + 0.777379 + +pittsburg-bridges-MATERIAL 0.864286 0.886147 + 0.867532 + 0.867532 + 0.877056 + +pittsburg-bridges-REL-L 0.695929 0.67619 - 0.705238 + 0.697619 + 0.69619 + +pittsburg-bridges-SPAN 0.68913 0.677193 - 0.709357 + 0.732164 + 0.687135 - +pittsburg-bridges-T-OR-D 0.87437 0.902381 + 0.902381 + 0.891905 + 0.882857 + +planning 0.725579 0.725525 - 0.747898 + 0.725676 + 0.736787 + +post-operative 0.711742 0.722222 + 0.722222 + 0.722222 + 0.711111 - +seeds 0.956303 0.961905 + 0.961905 + 0.942857 - 0.961905 + +statlog-australian-credit 0.678281 0.67971 + 0.7 + 0.686957 + 0.684058 + +statlog-german-credit 0.7562 0.762 + 0.777 + 0.769 + 0.762 + +statlog-heart 0.842299 0.848148 + 0.866667 + 0.840741 - 0.866667 + +statlog-image 0.976194 0.959307 - 0.962771 - 0.959307 - 0.962771 - +statlog-vehicle 0.800673 0.801413 + 0.817981 + 0.78607 - 0.79315 - +synthetic-control 0.990333 0.971667 - 0.995 + 0.981667 - 0.996667 + +tic-tac-toe 0.985385 0.987435 + 0.987435 + 0.98534 - 0.875758 - +vertebral-column-2clases 0.849153 0.829032 - 0.845161 - 0.83871 - 0.774194 - +wine 0.993281 0.977778 - 0.994444 + 0.988889 - 0.994444 + +zoo 0.960385 0.970476 + 0.98 + 0.96 - 0.97 + +we have better results 113 times +we have worse results 83 times +stree used 49 times better 25 times worse 24 times +odte used 49 times better 37 times worse 12 times +adaBoost used 49 times better 22 times worse 27 times +bagging used 49 times better 29 times worse 20 times \ No newline at end of file diff --git a/results/bagging.txt b/results/bagging.txt new file mode 100755 index 0000000..805d8fd --- /dev/null +++ b/results/bagging.txt @@ -0,0 +1,62 @@ +****************************************************************************************************** +* Best Hyperparameters found for datasets -- bagging classifier -- * +****************************************************************************************************** + +Date Time Type Classifier Dataset Nor Std Accuracy Reference +========== ======== ========== ========== ============================== === === ========= =========== +2020-12-19 18:36:31 gridsearch bagging balance-scale 1 0 0.8352000 0.9046280 - +2020-11-19 13:41:55 gridsearch bagging balloons 1 1 0.6833330 0.6625000 + +2020-11-20 3:48:59 gridsearch bagging breast-cancer-wisc-diag 1 1 0.9807020 0.9743450 + +2020-11-11 23:10:52 gridsearch bagging breast-cancer-wisc-prog 1 1 0.8337180 0.7993400 + +2020-11-23 12:10:11 gridsearch bagging breast-cancer-wisc 1 0 0.9771220 0.9702560 + +2020-11-19 23:44:41 gridsearch bagging breast-cancer 1 1 0.7519060 0.7382400 + +2020-11-16 13:37:13 gridsearch bagging cardiotocography-10clases 1 1 0.7022580 0.8277610 - +2020-11-16 13:35:23 gridsearch bagging cardiotocography-3clases 1 1 0.8490220 0.9201340 - +2020-12-12 13:27:08 gridsearch bagging conn-bench-sonar-mines-rocks 1 0 0.7304300 0.8336540 - +2020-12-20 1:53:07 gridsearch bagging cylinder-bands 1 0 0.6661340 0.7691410 - +2020-12-20 4:18:35 gridsearch bagging dermatology 1 0 0.9781560 0.9732780 + +2020-11-19 19:42:13 gridsearch bagging echocardiogram 1 1 0.8698010 0.8485270 + +2020-11-19 15:22:26 gridsearch bagging fertility 1 1 0.8900000 0.8840000 + +2020-11-20 7:04:39 gridsearch bagging haberman-survival 1 1 0.7386040 0.7392540 - +2020-11-23 15:28:21 gridsearch bagging heart-hungarian 1 0 0.8471070 0.8204750 + +2020-11-19 17:44:52 gridsearch bagging hepatitis 1 1 0.8645160 0.8232030 + +2020-12-21 7:18:24 gridsearch bagging ilpd-indian-liver 1 0 0.7307690 0.7150280 + +2020-11-20 7:39:16 gridsearch bagging ionosphere 1 1 0.9600810 0.9442150 + +2020-06-26 11:03:03 gridsearch bagging iris 1 0 0.9800000 0.9786560 + +2021-01-13 13:45:23 gridsearch bagging led-display 1 0 0.6950000 0.7102000 - +2021-01-14 0:28:42 gridsearch bagging libras 1 0 0.8194440 0.8911110 - +2020-12-24 16:13:21 gridsearch bagging low-res-spect 1 0 0.9133840 0.9028200 + +2020-11-20 1:24:35 gridsearch bagging lymphography 1 1 0.8919540 0.8554050 + +2020-12-22 11:05:56 gridsearch bagging mammographic 1 0 0.8376890 0.8274720 + +2020-11-20 2:26:50 gridsearch bagging molec-biol-promoter 1 1 0.9428570 0.8182690 + +2020-12-13 10:54:42 gridsearch bagging musk-1 1 0 0.8170390 0.8764710 - +2020-11-16 13:35:59 gridsearch bagging oocytes_merluccius_nucleus_4d 1 1 0.7925730 0.8399630 - +2020-11-16 13:35:27 gridsearch bagging oocytes_merluccius_states_2f 1 1 0.7299280 0.9299630 - +2020-11-16 13:35:46 gridsearch bagging oocytes_trisopterus_nucleus_2f 1 1 0.7279710 0.8333330 - +2020-11-16 13:35:41 gridsearch bagging oocytes_trisopterus_states_5b 1 1 0.8540800 0.9315790 - +2020-11-20 7:53:07 gridsearch bagging parkinsons 1 1 0.8615380 0.9202210 - +2020-12-23 22:40:26 gridsearch bagging pima 1 0 0.7773790 0.7671880 + +2020-11-21 14:57:13 gridsearch bagging pittsburg-bridges-MATERIAL 1 1 0.8770560 0.8642860 + +2020-11-20 12:30:28 gridsearch bagging pittsburg-bridges-REL-L 1 1 0.6961900 0.6959290 + +2020-11-16 13:35:56 gridsearch bagging pittsburg-bridges-SPAN 1 1 0.6871350 0.6891300 - +2020-11-16 13:35:54 gridsearch bagging pittsburg-bridges-T-OR-D 1 1 0.8828570 0.8743700 + +2020-11-22 0:19:48 gridsearch bagging planning 1 1 0.7367870 0.7255790 + +2020-11-16 13:35:58 gridsearch bagging post-operative 1 1 0.7111110 0.7117420 - +2020-11-20 17:28:36 gridsearch bagging seeds 1 1 0.9619050 0.9563030 + +2020-12-22 19:59:23 gridsearch bagging statlog-australian-credit 1 0 0.6840580 0.6782810 + +2020-12-24 0:02:12 gridsearch bagging statlog-german-credit 1 0 0.7620000 0.7562000 + +2020-11-20 22:17:22 gridsearch bagging statlog-heart 1 1 0.8666670 0.8422990 + +2020-11-16 14:04:13 gridsearch bagging statlog-image 1 1 0.9627710 0.9761940 - +2020-11-16 13:37:33 gridsearch bagging statlog-vehicle 1 1 0.7931500 0.8006730 - +2020-12-25 19:30:56 gridsearch bagging synthetic-control 1 0 0.9966670 0.9903330 + +2020-12-22 18:42:57 gridsearch bagging tic-tac-toe 1 0 0.8757580 0.9853850 - +2020-11-16 13:37:03 gridsearch bagging vertebral-column-2clases 1 1 0.7741940 0.8491530 - +2020-11-21 2:52:23 gridsearch bagging wine 1 1 0.9944440 0.9932810 + +2020-11-20 22:21:27 gridsearch bagging zoo 1 1 0.9700000 0.9603850 + +we have better results 29 times +we have worse results 20 times +we have equal results 0 times +stree used 0 times +bagging used 49 times +adaBoost used 0 times +odte used 0 times \ No newline at end of file diff --git a/results/odte.txt b/results/odte.txt new file mode 100755 index 0000000..e36716c --- /dev/null +++ b/results/odte.txt @@ -0,0 +1,62 @@ +****************************************************************************************************** +* Best Hyperparameters found for datasets -- odte classifier -- * +****************************************************************************************************** + +Date Time Type Classifier Dataset Nor Std Accuracy Reference +========== ======== ========== ========== ============================== === === ========= =========== +2020-12-23 7:47:46 gridsearch odte balance-scale 1 0 0.9712000 0.9046280 + +2021-01-11 11:52:54 gridsearch odte balloons 1 0 0.8666670 0.6625000 + +2020-12-16 22:24:05 gridsearch odte breast-cancer-wisc-diag 1 0 0.9824560 0.9743450 + +2020-12-16 20:10:07 gridsearch odte breast-cancer-wisc-prog 1 0 0.8487180 0.7993400 + +2020-11-22 14:14:12 gridsearch odte breast-cancer-wisc 1 0 0.9742650 0.9702560 + +2020-11-18 23:04:30 gridsearch odte breast-cancer 1 1 0.7449490 0.7382400 + +2020-12-07 23:56:19 gridsearch odte cardiotocography-10clases 1 0 0.7516510 0.8277610 - +2020-12-07 16:13:18 gridsearch odte cardiotocography-3clases 1 0 0.8612550 0.9201340 - +2020-12-21 13:58:11 gridsearch odte conn-bench-sonar-mines-rocks 1 0 0.7550520 0.8336540 - +2020-12-18 5:26:37 gridsearch odte cylinder-bands 1 0 0.6699600 0.7691410 - +2020-12-17 18:10:44 gridsearch odte dermatology 1 0 0.9864120 0.9732780 + +2020-11-22 5:37:43 gridsearch odte echocardiogram 1 0 0.8698010 0.8485270 + +2020-11-26 0:18:42 gridsearch odte fertility 1 0 0.8900000 0.8840000 + +2020-11-22 21:36:33 gridsearch odte haberman-survival 1 0 0.7745640 0.7392540 + +2020-11-22 18:24:35 gridsearch odte heart-hungarian 1 0 0.8437170 0.8204750 + +2020-11-22 4:30:37 gridsearch odte hepatitis 1 0 0.8774190 0.8232030 + +2020-12-22 9:29:28 gridsearch odte ilpd-indian-liver 1 0 0.7375770 0.7150280 + +2020-12-17 3:02:23 gridsearch odte ionosphere 1 0 0.9572230 0.9442150 + +2020-11-03 18:52:15 gridsearch odte iris 1 0 0.9933330 0.9786560 + +2020-12-07 18:41:58 gridsearch odte led-display 1 0 0.7310000 0.7102000 + +2021-01-09 4:17:20 gridsearch odte libras 1 0 0.8027780 0.8911110 - +2020-12-23 8:03:20 gridsearch odte low-res-spect 1 0 0.9134020 0.9028200 + +2020-11-22 6:38:17 gridsearch odte lymphography 1 0 0.8917240 0.8554050 + +2020-12-19 16:13:00 gridsearch odte mammographic 1 0 0.8428760 0.8274720 + +2020-12-12 12:15:28 gridsearch odte molec-biol-promoter 1 0 0.8961040 0.8182690 + +2020-12-23 4:28:26 gridsearch odte musk-1 1 0 0.8339030 0.8764710 - +2020-12-28 21:05:55 gridsearch odte oocytes_merluccius_nucleus_4d 1 0 0.8434190 0.8399630 + +2020-12-20 2:56:17 gridsearch odte oocytes_merluccius_states_2f 1 0 0.9295120 0.9299630 - +2020-12-08 19:38:34 gridsearch odte oocytes_trisopterus_nucleus_2f 1 0 0.7509640 0.8333330 - +2020-12-20 20:31:07 gridsearch odte oocytes_trisopterus_states_5b 1 0 0.8761360 0.9315790 - +2020-11-22 11:23:41 gridsearch odte parkinsons 1 0 0.8666670 0.9202210 - +2020-12-20 23:16:56 gridsearch odte pima 1 0 0.7890930 0.7671880 + +2020-11-22 13:27:02 gridsearch odte pittsburg-bridges-MATERIAL 1 0 0.8675320 0.8642860 + +2020-11-23 10:03:06 gridsearch odte pittsburg-bridges-REL-L 1 0 0.7052380 0.6959290 + +2020-11-22 13:03:57 gridsearch odte pittsburg-bridges-SPAN 1 0 0.7093570 0.6891300 + +2020-11-23 9:07:13 gridsearch odte pittsburg-bridges-T-OR-D 1 0 0.9023810 0.8743700 + +2020-11-23 3:03:50 gridsearch odte planning 1 0 0.7478980 0.7255790 + +2020-11-22 11:05:12 gridsearch odte post-operative 1 0 0.7222220 0.7117420 + +2020-11-22 16:26:56 gridsearch odte seeds 1 0 0.9619050 0.9563030 + +2020-12-19 17:15:01 gridsearch odte statlog-australian-credit 1 0 0.7000000 0.6782810 + +2020-12-21 0:36:07 gridsearch odte statlog-german-credit 1 0 0.7770000 0.7562000 + +2020-11-23 0:04:28 gridsearch odte statlog-heart 1 0 0.8666670 0.8422990 + +2020-12-08 21:15:04 gridsearch odte statlog-image 1 0 0.9627710 0.9761940 - +2020-12-08 21:36:41 gridsearch odte statlog-vehicle 1 0 0.8179810 0.8006730 + +2020-12-25 6:11:33 gridsearch odte synthetic-control 1 0 0.9950000 0.9903330 + +2020-12-19 2:22:48 gridsearch odte tic-tac-toe 1 0 0.9874350 0.9853850 + +2020-11-23 12:30:11 gridsearch odte vertebral-column-2clases 1 0 0.8451610 0.8491530 - +2020-11-23 1:18:25 gridsearch odte wine 1 0 0.9944440 0.9932810 + +2020-11-22 19:29:44 gridsearch odte zoo 1 0 0.9800000 0.9603850 + +we have better results 37 times +we have worse results 12 times +we have equal results 0 times +stree used 0 times +bagging used 0 times +adaBoost used 0 times +odte used 49 times \ No newline at end of file diff --git a/results/report.txt b/results/report.txt new file mode 100755 index 0000000..89d19b4 --- /dev/null +++ b/results/report.txt @@ -0,0 +1,62 @@ +****************************************************************************************************** +* Best Hyperparameters found for datasets -- any classifier -- * +****************************************************************************************************** + +Date Time Type Classifier Dataset Nor Std Accuracy Reference +========== ======== ========== ========== ============================== === === ========= =========== +2020-12-23 7:47:46 gridsearch odte balance-scale 1 0 0.9712000 0.9046280 + +2020-11-21 23:48:05 gridsearch adaBoost balloons 1 0 0.8833330 0.6625000 + +2020-12-16 22:24:05 gridsearch odte breast-cancer-wisc-diag 1 0 0.9824560 0.9743450 + +2020-12-16 20:10:07 gridsearch odte breast-cancer-wisc-prog 1 0 0.8487180 0.7993400 + +2020-11-23 12:10:11 gridsearch bagging breast-cancer-wisc 1 0 0.9771220 0.9702560 + +2020-11-19 23:44:41 gridsearch bagging breast-cancer 1 1 0.7519060 0.7382400 + +2020-12-07 23:56:19 gridsearch odte cardiotocography-10clases 1 0 0.7516510 0.8277610 - +2020-12-07 16:13:18 gridsearch odte cardiotocography-3clases 1 0 0.8612550 0.9201340 - +2020-12-21 13:58:11 gridsearch odte conn-bench-sonar-mines-rocks 1 0 0.7550520 0.8336540 - +2020-12-18 5:26:37 gridsearch odte cylinder-bands 1 0 0.6699600 0.7691410 - +2020-12-17 18:10:44 gridsearch odte dermatology 1 0 0.9864120 0.9732780 + +2020-11-22 5:37:43 gridsearch odte echocardiogram 1 0 0.8698010 0.8485270 + +2020-11-26 0:18:42 gridsearch odte fertility 1 0 0.8900000 0.8840000 + +2020-11-22 21:36:33 gridsearch odte haberman-survival 1 0 0.7745640 0.7392540 + +2020-11-23 15:28:21 gridsearch bagging heart-hungarian 1 0 0.8471070 0.8204750 + +2020-11-22 4:30:37 gridsearch odte hepatitis 1 0 0.8774190 0.8232030 + +2020-11-13 12:04:28 crossval stree ilpd-indian-liver 1 0 0.7426910 0.7150280 + +2020-11-20 7:39:16 gridsearch bagging ionosphere 1 1 0.9600810 0.9442150 + +2020-11-03 18:52:15 gridsearch odte iris 1 0 0.9933330 0.9786560 + +2020-12-07 18:41:58 gridsearch odte led-display 1 0 0.7310000 0.7102000 + +2021-01-14 0:28:42 gridsearch bagging libras 1 0 0.8194440 0.8911110 - +2020-12-23 8:03:20 gridsearch odte low-res-spect 1 0 0.9134020 0.9028200 + +2020-11-20 1:24:35 gridsearch bagging lymphography 1 1 0.8919540 0.8554050 + +2020-12-19 16:13:00 gridsearch odte mammographic 1 0 0.8428760 0.8274720 + +2020-11-20 2:26:50 gridsearch bagging molec-biol-promoter 1 1 0.9428570 0.8182690 + +2020-12-23 4:28:26 gridsearch odte musk-1 1 0 0.8339030 0.8764710 - +2020-12-28 21:05:55 gridsearch odte oocytes_merluccius_nucleus_4d 1 0 0.8434190 0.8399630 + +2020-12-20 2:56:17 gridsearch odte oocytes_merluccius_states_2f 1 0 0.9295120 0.9299630 - +2020-12-08 19:38:34 gridsearch odte oocytes_trisopterus_nucleus_2f 1 0 0.7509640 0.8333330 - +2020-12-20 20:31:07 gridsearch odte oocytes_trisopterus_states_5b 1 0 0.8761360 0.9315790 - +2020-11-22 11:23:41 gridsearch odte parkinsons 1 0 0.8666670 0.9202210 - +2020-12-20 23:16:56 gridsearch odte pima 1 0 0.7890930 0.7671880 + +2020-11-13 12:15:41 crossval stree pittsburg-bridges-MATERIAL 1 0 0.8861470 0.8642860 + +2020-11-23 10:03:06 gridsearch odte pittsburg-bridges-REL-L 1 0 0.7052380 0.6959290 + +2020-11-21 23:56:48 gridsearch adaBoost pittsburg-bridges-SPAN 1 0 0.7321640 0.6891300 + +2020-11-13 12:15:41 crossval stree pittsburg-bridges-T-OR-D 1 0 0.9023810 0.8743700 + +2020-11-23 3:03:50 gridsearch odte planning 1 0 0.7478980 0.7255790 + +2020-11-13 12:15:41 crossval stree post-operative 1 0 0.7222220 0.7117420 + +2020-11-13 12:15:41 crossval stree seeds 1 0 0.9619050 0.9563030 + +2020-12-19 17:15:01 gridsearch odte statlog-australian-credit 1 0 0.7000000 0.6782810 + +2020-12-21 0:36:07 gridsearch odte statlog-german-credit 1 0 0.7770000 0.7562000 + +2020-11-23 0:04:28 gridsearch odte statlog-heart 1 0 0.8666670 0.8422990 + +2020-12-08 21:15:04 gridsearch odte statlog-image 1 0 0.9627710 0.9761940 - +2020-12-08 21:36:41 gridsearch odte statlog-vehicle 1 0 0.8179810 0.8006730 + +2020-12-25 19:30:56 gridsearch bagging synthetic-control 1 0 0.9966670 0.9903330 + +2020-11-13 12:21:13 crossval stree tic-tac-toe 1 0 0.9874350 0.9853850 + +2020-11-23 12:30:11 gridsearch odte vertebral-column-2clases 1 0 0.8451610 0.8491530 - +2020-11-23 1:18:25 gridsearch odte wine 1 0 0.9944440 0.9932810 + +2020-11-22 19:29:44 gridsearch odte zoo 1 0 0.9800000 0.9603850 + +we have better results 37 times +we have worse results 12 times +we have equal results 0 times +stree used 6 times +bagging used 8 times +adaBoost used 2 times +odte used 33 times \ No newline at end of file diff --git a/results/stree.txt b/results/stree.txt new file mode 100755 index 0000000..b5e9d32 --- /dev/null +++ b/results/stree.txt @@ -0,0 +1,62 @@ +****************************************************************************************************** +* Best Hyperparameters found for datasets -- stree classifier -- * +****************************************************************************************************** + +Date Time Type Classifier Dataset Nor Std Accuracy Reference +========== ======== ========== ========== ============================== === === ========= =========== +2020-11-13 12:04:24 crossval stree balance-scale 1 0 0.9488000 0.9046280 + +2020-11-13 12:04:25 crossval stree balloons 1 0 0.8666670 0.6625000 + +2020-11-13 12:04:25 crossval stree breast-cancer-wisc-diag 1 0 0.9789320 0.9743450 + +2020-11-13 12:04:25 crossval stree breast-cancer-wisc-prog 1 0 0.8284620 0.7993400 + +2020-11-13 12:04:25 crossval stree breast-cancer-wisc 1 0 0.9656940 0.9702560 - +2020-11-13 12:04:26 crossval stree breast-cancer 1 0 0.7308530 0.7382400 - +2020-11-13 12:04:26 crossval stree cardiotocography-10clases 1 0 0.6665220 0.8277610 - +2020-11-13 12:04:27 crossval stree cardiotocography-3clases 1 0 0.8480740 0.9201340 - +2020-12-04 0:42:29 gridsearch stree conn-bench-sonar-mines-rocks 1 0 0.6882690 0.8336540 - +2020-11-13 12:04:27 crossval stree cylinder-bands 1 0 0.6326670 0.7691410 - +2020-11-13 12:04:27 crossval stree dermatology 1 0 0.9754540 0.9732780 + +2020-11-13 12:04:27 crossval stree echocardiogram 1 0 0.8472930 0.8485270 - +2020-11-13 12:04:27 crossval stree fertility 1 0 0.8800000 0.8840000 - +2020-11-13 12:04:28 crossval stree haberman-survival 1 0 0.7646750 0.7392540 + +2020-11-13 12:04:28 crossval stree heart-hungarian 1 0 0.8299240 0.8204750 + +2020-11-13 12:04:28 crossval stree hepatitis 1 0 0.8645160 0.8232030 + +2020-11-13 12:04:28 crossval stree ilpd-indian-liver 1 0 0.7426910 0.7150280 + +2020-11-13 12:04:28 crossval stree ionosphere 1 0 0.9487320 0.9442150 + +2020-11-13 12:04:28 crossval stree iris 0 0 0.9800000 0.9786560 + +2020-11-13 12:15:39 crossval stree led-display 1 0 0.7120000 0.7102000 + +2020-12-10 1:28:06 gridsearch stree libras 1 0 0.7027780 0.8911110 - +2020-12-09 23:13:34 gridsearch stree low-res-spect 1 0 0.8794920 0.9028200 - +2020-11-13 12:15:39 crossval stree lymphography 1 0 0.8648280 0.8554050 + +2020-11-13 12:15:39 crossval stree mammographic 1 0 0.8293720 0.8274720 + +2020-12-03 19:16:25 gridsearch stree molec-biol-promoter 1 0 0.8108220 0.8182690 - +2020-12-11 12:21:33 gridsearch stree musk-1 1 0 0.7899120 0.8764710 - +2020-11-13 12:15:39 crossval stree oocytes_merluccius_nucleus_4d 1 0 0.8082210 0.8399630 - +2020-11-13 12:15:39 crossval stree oocytes_merluccius_states_2f 1 0 0.9119030 0.9299630 - +2020-11-13 12:15:40 crossval stree oocytes_trisopterus_nucleus_2f 1 0 0.7476910 0.8333330 - +2020-11-13 12:15:40 crossval stree oocytes_trisopterus_states_5b 1 0 0.8453610 0.9315790 - +2020-11-13 12:15:40 crossval stree parkinsons 1 0 0.8461540 0.9202210 - +2020-11-13 12:15:41 crossval stree pima 1 0 0.7800020 0.7671880 + +2020-11-13 12:15:41 crossval stree pittsburg-bridges-MATERIAL 1 0 0.8861470 0.8642860 + +2020-11-13 12:15:41 crossval stree pittsburg-bridges-REL-L 1 0 0.6761900 0.6959290 - +2020-11-13 12:15:41 crossval stree pittsburg-bridges-SPAN 1 0 0.6771930 0.6891300 - +2020-11-13 12:15:41 crossval stree pittsburg-bridges-T-OR-D 1 0 0.9023810 0.8743700 + +2020-11-13 12:15:41 crossval stree planning 1 0 0.7255250 0.7255790 - +2020-11-13 12:15:41 crossval stree post-operative 1 0 0.7222220 0.7117420 + +2020-11-13 12:15:41 crossval stree seeds 1 0 0.9619050 0.9563030 + +2020-11-13 12:15:41 crossval stree statlog-australian-credit 1 0 0.6797100 0.6782810 + +2020-11-13 12:21:08 crossval stree statlog-german-credit 1 0 0.7620000 0.7562000 + +2020-11-13 12:21:08 crossval stree statlog-heart 1 0 0.8481480 0.8422990 + +2020-11-13 12:21:12 crossval stree statlog-image 1 0 0.9593070 0.9761940 - +2020-11-13 12:21:13 crossval stree statlog-vehicle 1 0 0.8014130 0.8006730 + +2020-12-07 10:42:19 gridsearch stree synthetic-control 1 0 0.9716670 0.9903330 - +2020-11-13 12:21:13 crossval stree tic-tac-toe 1 0 0.9874350 0.9853850 + +2020-11-13 12:21:14 crossval stree vertebral-column-2clases 1 0 0.8290320 0.8491530 - +2020-11-13 12:21:14 crossval stree wine 1 0 0.9777780 0.9932810 - +2020-11-13 12:21:14 crossval stree zoo 1 0 0.9704760 0.9603850 + +we have better results 25 times +we have worse results 24 times +we have equal results 0 times +stree used 49 times +bagging used 0 times +adaBoost used 0 times +odte used 0 times \ No newline at end of file diff --git a/test.ipynb b/test.ipynb index 2abaca0..010b543 100644 --- a/test.ipynb +++ b/test.ipynb @@ -57,7 +57,6 @@ "metadata": {}, "outputs": [ { - "output_type": "execute_result", "data": { "text/plain": [ "BaggingClassifier(base_estimator=Stree(C=17, random_state=1,\n", @@ -66,8 +65,9 @@ " random_state=1)" ] }, + "execution_count": 11, "metadata": {}, - "execution_count": 11 + "output_type": "execute_result" } ], "source": [ @@ -89,14 +89,14 @@ "metadata": {}, "outputs": [ { - "output_type": "execute_result", "data": { "text/plain": [ "0.96" ] }, + "execution_count": 13, "metadata": {}, - "execution_count": 13 + "output_type": "execute_result" } ], "source": [ @@ -109,14 +109,14 @@ "metadata": {}, "outputs": [ { - "output_type": "execute_result", "data": { "text/plain": [ "0.9866666666666667" ] }, + "execution_count": 14, "metadata": {}, - "execution_count": 14 + "output_type": "execute_result" } ], "source": [ @@ -708,9 +708,9 @@ ], "metadata": { "kernelspec": { - "display_name": "Python 3.7.6 64-bit ('general': venv)", + "display_name": "Python 3", "language": "python", - "name": "python37664bitgeneralvenvfbd0a23e74cf4e778460f5ffc6761f39" + "name": "python3" }, "language_info": { "codemirror_mode": { @@ -722,9 +722,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.4-final" + "version": "3.8.3" } }, "nbformat": 4, "nbformat_minor": 4 -} \ No newline at end of file +} diff --git a/test_validation.ipynb b/test_validation.ipynb index 178b15c..8b22de9 100644 --- a/test_validation.ipynb +++ b/test_validation.ipynb @@ -805,7 +805,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.2" + "version": "3.8.3" } }, "nbformat": 4,