rmontanana/BayesNet
rmontanana/BayesNet
1
0
Fork 0
Code Issues 2 Pull Requests Packages Projects Releases 3 Wiki Activity

Complete Folding Test

Browse Source
This commit is contained in:
Ricardo Montañana Gómez 2023-10-07 01:23:36 +02:00
parent 1287160c47
commit 8c3864f3c8
Signed by: rmontanana
GPG Key ID: 46064262FD9A7ADE
5 changed files with 26 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
Makefile
View File

@ -15,10 +15,7 @@ define ClearTests
rm -f $(f_debug)/tests/$$t ; \
fi ; \
done
$(eval nfiles=$(find . -name "*.gcda" -print))
@if test "${nfiles}" != "" ; then \
find . -name "*.gcda" -print0 | xargs -0 rm 2>/dev/null ;\
fi ;
@find . -name "*.gcda" -print0 | xargs -0 rm 2>/dev/null ;
endef
@ -106,8 +103,8 @@ testb: ## Run BayesNet tests (opt="-s") to verbose output the tests, (opt="-c='T
coverage: ## Run tests and generate coverage report (build/index.html)
@echo ">>> Building tests with coverage...";
@$(MAKE) test
@cd $(f_debug) ;
@gcovr --config ../gcovr.cfg ;
@cd $(f_debug) ; \
gcovr --config ../gcovr.cfg tests ;
@echo ">>> Done";

13
src/Platform/Folding.cc
View File

@ -47,7 +47,7 @@ namespace platform {
{
stratified_indices = vector<vector<int>>(k);
int fold_size = n / k;
cout << "Fold SIZE: " << fold_size << endl;
// Compute class counts and indices
auto class_indices = map<int, vector<int>>();
vector<int> class_counts(*max_element(y.begin(), y.end()) + 1, 0);
@ -61,11 +61,14 @@ namespace platform {
}
// Assign indices to folds
for (auto label = 0; label < class_counts.size(); ++label) {
auto num_samples_to_take = class_counts[label] / k;
if (num_samples_to_take == 0)
auto num_samples_to_take = class_counts.at(label) / k;
if (num_samples_to_take == 0) {
cerr << "Warning! The number of samples in class " << label << " (" << class_counts.at(label)
<< ") is less than the number of folds (" << k << ")." << endl;
faulty = true;
continue;
}
auto remainder_samples_to_take = class_counts[label] % k;
cout << "Remainder samples to take: " << remainder_samples_to_take << endl;
for (auto fold = 0; fold < k; ++fold) {
auto it = next(class_indices[label].begin(), num_samples_to_take);
move(class_indices[label].begin(), it, back_inserter(stratified_indices[fold])); // ##
@ -74,12 +77,10 @@ namespace platform {
auto chosen = vector<bool>(k, false);
while (remainder_samples_to_take > 0) {
int fold = (rand() % static_cast<int>(k));
cout << "-candidate: " << fold << endl;
if (chosen.at(fold)) {
continue;
}
chosen[fold] = true;
cout << "One goes to fold " << fold << " that had " << stratified_indices[fold].size() << " elements before" << endl;
auto it = next(class_indices[label].begin(), 1);
stratified_indices[fold].push_back(*class_indices[label].begin());
class_indices[label].erase(class_indices[label].begin(), it);

2
src/Platform/Folding.h
View File

@ -29,10 +29,12 @@ namespace platform {
vector<int> y;
vector<vector<int>> stratified_indices;
void build();
bool faulty = false; // Only true if the number of samples of any class is less than the number of folds.
public:
StratifiedKFold(int k, const vector<int>& y, int seed = -1);
StratifiedKFold(int k, torch::Tensor& y, int seed = -1);
pair<vector<int>, vector<int>> getFold(int nFold) override;
bool isFaulty() { return faulty; }
};
}
#endif

3
tests/TestBayesMetrics.cc
View File

@ -22,7 +22,8 @@ TEST_CASE("Metrics Test", "[BayesNet]")
{"diabetes", 0.0345470614}
};
map<string, vector<pair<int, int>>> resultsMST = {
{"glass", {{0,6}, {0,5}, {0,3}, {6,2}, {6,7}, {5,1}, {5,8}, {5,4}}},
//{"glass", {{0,6}, {0,5}, {0,3}, {6,2}, {6,7}, {5,1}, {5,8}, {5,4}}},
{"glass", {{0,6}, {0,5}, {0,3}, {5,1}, {5,8}, {5,4}, {6,2}, {6,7}}},
{"iris", {{0,1},{0,2},{1,3}}},
{"ecoli", {{0,1}, {0,2}, {1,5}, {1,3}, {5,6}, {5,4}}},
{"diabetes", {{0,7}, {0,2}, {0,6}, {2,3}, {3,4}, {3,5}, {4,1}}}

23
tests/TestFolding.cc
View File

@ -66,27 +66,28 @@ TEST_CASE("StratifiedKFold Test", "[Platform][StratifiedKFold]")
auto [train_indicesv, test_indicesv] = stratified_kfoldv.getFold(fold);
REQUIRE(train_indicest == train_indicesv);
REQUIRE(test_indicest == test_indicesv);
bool result = train_indicest.size() == number || train_indicest.size() == number + 1;
REQUIRE(result);
REQUIRE(train_indicest.size() + test_indicest.size() == raw.nSamples);
// In the worst case scenario, the number of samples in the training set is number + raw.classNumStates
// because in that fold can come one remainder sample from each class.
REQUIRE(train_indicest.size() <= number + raw.classNumStates);
// If the number of samples in any class is less than the number of folds, then the fold is faulty.
// and the number of samples in the training set + test set will be less than nSamples
if (!stratified_kfoldt.isFaulty()) {
REQUIRE(train_indicest.size() + test_indicest.size() == raw.nSamples);
} else {
REQUIRE(train_indicest.size() + test_indicest.size() <= raw.nSamples);
}
auto train_t = torch::tensor(train_indicest);
auto ytrain = raw.yt.index({ train_t });
cout << "dataset=" << file_name << endl;
cout << "nSamples=" << raw.nSamples << endl;;
cout << "number=" << number << endl;
cout << "train_indices.size()=" << train_indicest.size() << endl;
cout << "test_indices.size()=" << test_indicest.size() << endl;
cout << "Class Name = " << raw.classNamet << endl;
// Check that the class labels have been equally assign to each fold
for (const auto& idx : train_indicest) {
counts[fold][ytrain[idx].item<int>()]++;
counts[fold][raw.yt[idx].item<int>()]++;
}
}
// Test the fold counting of every class
for (int fold = 0; fold < nFolds; ++fold) {
for (int j = 1; j < nFolds - 1; ++j) {
for (int k = 0; k < raw.classNumStates; ++k) {
REQUIRE(abs(counts.at(fold).at(k) - counts.at(fold).at(j)) <= 1);
REQUIRE(abs(counts.at(fold).at(k) - counts.at(j).at(k)) <= 1);
}
}
}