private static int HighestGainAttribute(DataBag data)
{
var tableList = FrequencyTable.GetFrequencyTables(data.dataList);
if (tableList == null)
{
throw new ArgumentNullException(nameof(tableList));
}
var attribCount = DataSet.Attributes.Count - 1;
var highestGain = new KeyValuePair <int, double>();
var sum = tableList.First().AllRowsLeft;
var qualifierCount = tableList[0].QualifierCount;
var entropy = Entropy(sum, qualifierCount);
for (var i = 0; i < attribCount; i++)
{
var currGain = Gain(entropy, tableList[i]);
if (highestGain.Value < currGain)
{
highestGain = new KeyValuePair <int, double>(i, currGain);
}
}
return(highestGain.Key);
}
private static double Gain(double entropy, FrequencyTable table)
{
var allInstancesLeftCount = table.AllRowsLeft;
var attributeValueCount = DataSet.Attributes[table.AttributeIndex].ValueCount;
for (var value = 0; value < attributeValueCount; value++)
{
var qualifierCount = table.ValueQualifierSum(value);
var attributesRowAmount = table.AttributeRowCount(value);
entropy -= ((double)attributesRowAmount / allInstancesLeftCount) * Entropy(attributesRowAmount, qualifierCount);
}
return(entropy);
}
