/// <summary>
/// Invoke weka jar to create an arff from a csv file
/// </summary>
protected override RandomForestFromWekaOutput Perform(RandomForestFromWekaInput input)
{
s_logger.Debug($"RandomForestFromWeka starts");
try
{
var outputTreeFile = $"{input.TrainingSetCsv}.wtree";
var outputArffFile = $"{input.TrainingSetCsv}.arff";
RandomForest forest = null;
if (!File.Exists(outputTreeFile))
{
// lets analyze this in here
try
{
var arffExitCode = WriteArff(input, outputArffFile);
s_logger.Debug($"ArffWrite returns exitCode={arffExitCode}");
if (File.Exists(outputArffFile))
{
var exitCode = RunWeka(input, outputArffFile, outputTreeFile);
s_logger.Debug($"RandomForestFromWeka returns exitCode={exitCode}");
}
else
{
s_logger.Error($"Could not write arff file for [{input.TrainingSetCsv}]");
}
}
catch (Exception e)
{
s_logger.Error(e, "An exception ocurred when writting forest from weka");
}
finally
{
if (!File.Exists(outputTreeFile))
{
throw new Exception($"RandomForestFromWeka task failed to write output to [{outputTreeFile}]");
}
}
}
// and now load the forest
forest = RandomForest.FromWekaFile(outputTreeFile);
// done
return(new RandomForestFromWekaOutput(forest, input.TrainingSetCsv, outputTreeFile));
}
finally
{
s_logger.Debug($"RandomForestFromWeka ends in {Stopwatch.ElapsedMilliseconds}ms");
}
}
private static void EvaluateForests(Args arguments)
{
Contract.Requires(arguments.InputDirectory != null, "You must specify an input directory");
Contract.Requires(Directory.Exists(arguments.InputDirectory), "The input directory must exist");
// we have a bunch of forests and we will compare them by classifying against
// unknown samples. The forests are in weka format (wtree)
s_logger.Info("Evaluating random forests...");
// create the tree from the sample csvs...
var treeEvaluationBlock = new ActionBlock <string>(i =>
{
var classifier = RandomForest.FromWekaFile(i);
s_logger.Info($"Evaluating forest from [{i}] against universe");
// and now evaluate
foreach (var evaluationFile in Directory.EnumerateFiles(arguments.InputDirectory, "*.csv"))
{
if (!evaluationFile.Contains("-sample"))
{
classifier.EvaluateOnTrainingSet(evaluationFile, true, false, 0);
}
}
},
new ExecutionDataflowBlockOptions()
{
MaxDegreeOfParallelism = 1
}
);
// post each action
foreach (var treeFile in Directory.EnumerateFiles(arguments.InputDirectory, "*.wtree"))
{
treeEvaluationBlock.Post(treeFile);
}
// complete
treeEvaluationBlock.Complete();
// wait
treeEvaluationBlock.Completion.Wait();
// done...
}