Skip to content

KMeans: move empty clusters warning from algorithm to client #336

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
weefuzzy merged 2 commits into from
Dec 12, 2025
Merged

KMeans: move empty clusters warning from algorithm to client #336

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a6a8f6f
KMeans: move empty clusters warning from algorithm to client
weefuzzy Dec 11, 2025
1945f63
added the warning mechanism to skmean
tremblap Dec 12, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion include/flucoma/algorithms/public/KMeans.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -232,6 +232,7 @@ class KMeans
index size() const { return mMeans.rows(); }
index getK() const { return mMeans.rows(); }
index nAssigned() const { return mAssignments.size(); }
index nEmpty() const { return std::count(mEmpty.begin(), mEmpty.end(), true); }

void getAssignments(FluidTensorView<index, 1> out) const
{
Expand Down Expand Up @@ -295,7 +296,6 @@ class KMeans
}
if (kAssignment.size() == 0)
{
std::cout << "Warning: empty cluster" << std::endl;
mEmpty[asUnsigned(k)] = true;
return;
}
Expand Down
43 changes: 28 additions & 15 deletions include/flucoma/clients/nrt/KMeansClient.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -85,11 +85,9 @@ class KMeansClient : public FluidBaseClient,
if (dataSet.size() == 0) return Error<IndexVector>(EmptyDataSet);
if (k <= 1) return Error<IndexVector>(SmallK);
if(mTracker.changed(k)) mAlgorithm.clear();
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
return getCounts(assignments, k);
auto [result, _] =
train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
return result;
}

MessageResult<IndexVector> fitPredict(InputDataSetClientRef datasetClient,
Expand All @@ -106,13 +104,11 @@ class KMeansClient : public FluidBaseClient,
if (k <= 1) return Error<IndexVector>(SmallK);
if (maxIter <= 0) maxIter = 100;
if(mTracker.changed(k)) mAlgorithm.clear();
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto [result, assignments] =
train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
StringVectorView ids = dataSet.getIds();
labelsetClientPtr->setLabelSet(getLabels(ids, assignments));
return getCounts(assignments, k);
return result;
}

MessageResult<IndexVector> predict(InputDataSetClientRef datasetClient,
Expand Down Expand Up @@ -175,12 +171,10 @@ class KMeansClient : public FluidBaseClient,
if (dataSet.size() == 0) return Error<IndexVector>(EmptyDataSet);
if (k <= 1) return Error<IndexVector>(SmallK);
if (maxIter <= 0) maxIter = 100;
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto [result, _] =
train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
transform(srcClient, dstClient);
return getCounts(assignments, k);
return result;
}

MessageResult<index> predictPoint(InputBufferPtr data) const
Expand Down Expand Up @@ -263,6 +257,25 @@ class KMeansClient : public FluidBaseClient,


private:
using DataSet = FluidDataSet<std::string, double, 1>;

std::pair<MessageResult<IndexVector>, IndexVector>
train(DataSet const& dataSet, index k, index maxIter , index initMethod, index randomSeed)
{
mAlgorithm.train(
dataSet, k, maxIter,
static_cast<algorithm::KMeans::InitMethod>(initMethod), randomSeed);
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto training_result = MessageResult<IndexVector>(getCounts(assignments,k));
if(mAlgorithm.nEmpty() > 0)
{
training_result.set(Result::Status::kWarning);
training_result.addMessage("There were empty clusters; perhaps numClusters is too high.");
}
return {training_result, assignments};
}

IndexVector getCounts(IndexVector assignments, index k) const
{
IndexVector counts(k);
Expand Down
40 changes: 25 additions & 15 deletions include/flucoma/clients/nrt/SKMeansClient.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -84,11 +84,8 @@ class SKMeansClient : public FluidBaseClient,
if (dataSet.size() == 0) return Error<IndexVector>(EmptyDataSet);
if (k <= 1) return Error<IndexVector>(SmallK);
if(mTracker.changed(k)) mAlgorithm.clear();
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
return getCounts(assignments, k);
auto [result, _] = train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
return result;
}


Expand All @@ -106,13 +103,10 @@ class SKMeansClient : public FluidBaseClient,
if (k <= 1) return Error<IndexVector>(SmallK);
if (maxIter <= 0) maxIter = 100;
if(mTracker.changed(k)) mAlgorithm.clear();
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto [result, assignments] = train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
StringVectorView ids = dataSet.getIds();
labelsetClientPtr->setLabelSet(getLabels(ids, assignments));
return getCounts(assignments, k);
return result;
}


Expand Down Expand Up @@ -178,12 +172,9 @@ class SKMeansClient : public FluidBaseClient,
if (k <= 1) return Error<IndexVector>(SmallK);
if (maxIter <= 0) maxIter = 100;
if(mTracker.changed(k)) mAlgorithm.clear();
mAlgorithm.train(dataSet, k, maxIter, static_cast<InitMethod>(get<kInit>()),
get<kRandomSeed>());
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto [result, _] = train(dataSet, k, maxIter, get<kInit>(), get<kRandomSeed>());
encode(srcClient, dstClient);
return getCounts(assignments, k);
return result;
}

MessageResult<index> predictPoint(BufferPtr data) const
Expand Down Expand Up @@ -266,6 +257,25 @@ class SKMeansClient : public FluidBaseClient,


private:
using DataSet = FluidDataSet<std::string, double, 1>;

std::pair<MessageResult<IndexVector>, IndexVector>
train(DataSet const& dataSet, index k, index maxIter , index initMethod, index randomSeed)
{
mAlgorithm.train(
dataSet, k, maxIter,
static_cast<algorithm::KMeans::InitMethod>(initMethod), randomSeed);
IndexVector assignments(dataSet.size());
mAlgorithm.getAssignments(assignments);
auto training_result = MessageResult<IndexVector>(getCounts(assignments,k));
if(mAlgorithm.nEmpty() > 0)
{
training_result.set(Result::Status::kWarning);
training_result.addMessage("There were empty clusters; perhaps numClusters is too high.");
}
return {training_result, assignments};
}

IndexVector getCounts(IndexVector assignments, index k) const
{
IndexVector counts(k);
Expand Down