public IDataTableAnalysis GetAnalysis()
{
if (_analysis == null)
{
var analysis = new DataTableAnalysis(this);
Process(analysis);
_analysis = analysis;
}
return(_analysis);
}
public static NaiveBayes Train(IDataTable table)
{
// analyse the table to get the set of class values
var classColumnIndex = table.TargetColumnIndex;
var analysis = new DataTableAnalysis(table, classColumnIndex);
table.Process(analysis);
var classInfo = analysis.ColumnInfo.Single();
if (classInfo.DistinctValues == null)
{
throw new Exception("Too many class values");
}
// analyse the data per class
var classBasedFrequency = classInfo.DistinctValues.Select(cv => Tuple.Create <string, IRowProcessor>(cv.ToString(), new FrequencyAnalysis(table, classColumnIndex)));
var frequencyAnalysis = new ClassificationBasedRowProcessor(classBasedFrequency, classColumnIndex);
table.Process(frequencyAnalysis);
// create the per-class summaries from the frequency table
var classList = new List <NaiveBayes.ClassSummary>();
foreach (var classSummary in frequencyAnalysis.All)
{
var classLabel = classSummary.Item1;
var frequency = (FrequencyAnalysis)classSummary.Item2;
var columnList = new List <NaiveBayes.Column>();
foreach (var column in frequency.ColumnInfo)
{
var continuous = column as NumberCollector;
if (column is FrequencyCollector categorical)
{
var total = (double)categorical.Total;
if (total > 0)
{
var list = new List <NaiveBayes.CategorialProbability>();
foreach (var item in categorical.Frequency)
{
var categoryProbability = item.Value / total;
list.Add(new NaiveBayes.CategorialProbability {
Category = item.Key,
LogProbability = Math.Log(categoryProbability),
Probability = categoryProbability
});
}
columnList.Add(new NaiveBayes.Column {
Type = NaiveBayes.ColumnType.Categorical,
ColumnIndex = categorical.ColumnIndex,
Probability = list
});
}
}
else
{
var variance = continuous?.Variance;
if (variance != null)
{
var mean = continuous.Mean;
columnList.Add(new NaiveBayes.Column {
Type = NaiveBayes.ColumnType.ContinuousGaussian,
ColumnIndex = continuous.ColumnIndex,
Mean = mean,
Variance = variance.Value
});
}
}
}
var probability = frequencyAnalysis.GetProbability(classLabel);
classList.Add(new NaiveBayes.ClassSummary {
Label = classLabel,
ColumnSummary = columnList,
LogPrior = Math.Log(probability),
Prior = probability
});
}
return(new NaiveBayes {
Class = classList
});
}