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);
}
public static void ID3(Node n, DataBag data, int level = 1)
{
if (DataSet.MaxLevel < level)
{
DataSet.MaxLevel = level;
}
n.IsLeaf = data.IsAtomic;
if (n.IsLeaf)
{
n.Qualifier = data.HighestQualifierCount();
return;
}
n.Attribute = HighestGainAttribute(data);
for (var attributeValue = 0; attributeValue < DataSet.Attributes[n.Attribute].ValueCount; attributeValue++)
{
var newNode = new Node(n.Attribute)
{
OriginEdge = attributeValue
};
var newData = new DataBag(data, n.Attribute, attributeValue);
n.Paths.Add(newNode);
ID3(newNode, newData, level + 1);
}
}
public DataBag(DataBag oldBag, int attribute, int value)
{
dataList = oldBag.dataList.Where(instance => instance.Data[attribute] == value).ToList();
AttributesLeft = oldBag.AttributesLeft - 1;
}