rmontanana/mdlp
1
1
Fork 0
mirror of https://github.com/rmontanana/mdlp.git synced 2025-08-17 08:25:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

Discretizer (#8)

Browse Source 
* Add better check in testKBins.py

* Add Discretizer base class for Both discretizers

* Refactor order of constructors init
This commit is contained in:
Ricardo Montañana Gómez
2024-06-05 17:53:08 +02:00
committed by GitHub
parent f258fc220f
commit 638bb2a59e
12 changed files with 294 additions and 154 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
BinDisc.cpp
View File

@@ -7,7 +7,8 @@
namespace mdlp {
BinDisc::BinDisc(int n_bins, strategy_t strategy) : n_bins{ n_bins }, strategy{ strategy }
BinDisc::BinDisc(int n_bins, strategy_t strategy) :
Discretizer(), n_bins{ n_bins }, strategy{ strategy }
{
if (n_bins < 3) {
throw std::invalid_argument("n_bins must be greater than 2");
@@ -16,6 +17,7 @@ namespace mdlp {
BinDisc::~BinDisc() = default;
void BinDisc::fit(samples_t& X)
{
// y is included for compatibility with the Discretizer interface
cutPoints.clear();
if (X.empty()) {
cutPoints.push_back(std::numeric_limits<precision_t>::max());
@@ -27,6 +29,10 @@ namespace mdlp {
fit_uniform(X);
}
}
void BinDisc::fit(samples_t& X, labels_t& y)
{
fit(X);
}
std::vector<precision_t> linspace(precision_t start, precision_t end, int num)
{
// Doesn't include end point as it is not needed
@@ -90,49 +96,4 @@ namespace mdlp {
// Remove first as it is not needed
cutPoints.erase(cutPoints.begin());
}
labels_t& BinDisc::transform(const samples_t& X)
{
discretizedData.clear();
discretizedData.reserve(X.size());
for (const precision_t& item : X) {
auto upper = std::upper_bound(cutPoints.begin(), cutPoints.end(), item);
discretizedData.push_back(upper - cutPoints.begin());
}
return discretizedData;
}
}
// void BinDisc::fit_quantile(samples_t& X)
// {
// cutPoints.clear();
// if (X.empty()) {
// cutPoints.push_back(std::numeric_limits<float>::max());
// return;
// }
// samples_t data = X;
// std::sort(data.begin(), data.end());
// float min_val = data.front();
// float max_val = data.back();
// // Handle case of all data points having the same value
// if (min_val == max_val) {
// cutPoints.push_back(std::numeric_limits<float>::max());
// return;
// }
// int first = X.size() / n_bins;
// cutPoints.push_back(data.at(first - 1));
// int bins_done = 1;
// int prev = first - 1;
// while (bins_done < n_bins) {
// int next = first * (bins_done + 1) - 1;
// while (next < X.size() && data.at(next) == data[prev]) {
// ++next;
// }
// if (next == X.size() || bins_done == n_bins - 1) {
// cutPoints.push_back(std::numeric_limits<float>::max());
// break;
// } else {
// cutPoints.push_back(data[next]);
// bins_done++;
// prev = next;
// }
// }
// }
}