2 files changed, 12 insertions, 1 deletions
diff --git a/ml/kmeans/SamplesBuffer.cc b/ml/kmeans/SamplesBuffer.cc
index f8211fb54..d276c6e09 100644
--- a/ml/kmeans/SamplesBuffer.cc
+++ b/ml/kmeans/SamplesBuffer.cc
@@ -130,7 +130,13 @@ std::vector<DSample> SamplesBuffer::preprocess() {
     DSamples.reserve(OutN);
     Preprocessed = true;
 
+    uint32_t MaxMT = std::numeric_limits<uint32_t>::max();
+    uint32_t CutOff = static_cast<double>(MaxMT) * SamplingRatio;
+
     for (size_t Idx = NumSamples - OutN; Idx != NumSamples; Idx++) {
+        if (RandNums[Idx] > CutOff)
+            continue;
+
         DSample DS;
         DS.set_size(NumDimsPerSample * (LagN + 1));
 
diff --git a/ml/kmeans/SamplesBuffer.h b/ml/kmeans/SamplesBuffer.h
index 5c3cb1a97..1c7215cca 100644
--- a/ml/kmeans/SamplesBuffer.h
+++ b/ml/kmeans/SamplesBuffer.h
@@ -80,9 +80,11 @@ class SamplesBuffer {
 public:
     SamplesBuffer(CalculatedNumber *CNs,
                   size_t NumSamples, size_t NumDimsPerSample,
-                  size_t DiffN = 1, size_t SmoothN = 3, size_t LagN = 3) :
+                  size_t DiffN, size_t SmoothN, size_t LagN,
+                  double SamplingRatio, std::vector<uint32_t> &RandNums) :
         CNs(CNs), NumSamples(NumSamples), NumDimsPerSample(NumDimsPerSample),
         DiffN(DiffN), SmoothN(SmoothN), LagN(LagN),
+        SamplingRatio(SamplingRatio), RandNums(RandNums),
         BytesPerSample(NumDimsPerSample * sizeof(CalculatedNumber)),
         Preprocessed(false) {};
 
@@ -129,6 +131,9 @@ private:
     size_t DiffN;
     size_t SmoothN;
     size_t LagN;
+    double SamplingRatio;
+    std::vector<uint32_t> &RandNums;
+
     size_t BytesPerSample;
     bool Preprocessed;
 };