private void CalculateColumnInformationGain()
{
foreach (var decisionColumn in DecisionColumns)
{
decisionColumn.InformationGain = TotalEntropy - decisionColumn.Entropy;
decisionColumn.ColumnIndex = DecisionColumns.IndexOf(decisionColumn);
}
}
private void DoUnknownThing(List <int> bestOneIndexes, List <List <int> > fullHierarchy)
{
var index = -1;
var selectedIndexesForNextRound = new List <int>();
var newHierarchy = new List <List <int> >();
var locallyUsedColumnIndexes = new List <int>();
var localAttributeValue = string.Empty;
foreach (var bestOneIndex in bestOneIndexes)
{
index++;
var bestOne = DecisionColumns.First(col => col.ColumnIndex == bestOneIndex);
bestOne.ColumnIndex = bestOneIndex;
var rejectedList =
DecisionColumns.Where(column => AlreadyUsedIndexes.Contains((int)column.ColumnIndex));
AlreadyUsedIndexes.Add((int)bestOne.ColumnIndex); //Dodajemy ja do listy na przyszlosc
Console.WriteLine($"Columnd index {bestOne.ColumnIndex}");
foreach (var attribute in bestOne.Attributes) // i dla kazdego atrybutu najlepszej kolumny czyli np. dla pogody iterujemy sie po: sloecznie, pochmurno, itd.
{
var attributeRows =
attribute.NegativeRowNumbers.Concat(attribute.PositiveRowNumbers)
.ToList(); // bierzemy wszystkie indeksy wierszy w jakich wystepuja
rejectedList = DecisionColumns.Where(column => AlreadyUsedIndexes.Contains((int)column.ColumnIndex));
var tempList = DecisionColumns.Except(rejectedList);
foreach (var decisionColumn in tempList
) // i iterujemy sie po wszystkich niewykorzystanych kolumnach
{
decisionColumn.ColumnIndex = DecisionColumns.IndexOf(decisionColumn);
var innerAttributes = decisionColumn.Attributes
.Where(innerAttribute =>
innerAttribute.PositiveRowNumbers.Intersect(attributeRows).Any() ||
innerAttribute.NegativeRowNumbers.Intersect(attributeRows).Any()).ToList();
//to sa te atrybuty, ktory wystepuja w wierszach 'attribute'
decisionColumn.CaseEntropy = 0;
var allNegatives = innerAttributes.SelectMany(innerAttr => innerAttr.NegativeRowNumbers)
.ToList();
if (allNegatives.Count(neg => attributeRows.Contains(neg)) > 0)
{
foreach (var innerAttribute in innerAttributes)
{
var innerRows = innerAttribute.NegativeRowNumbers
.Concat(innerAttribute.PositiveRowNumbers)
.ToList();
var innerPositiveRows = innerAttribute.PositiveRowNumbers
.Where(inner => attributeRows.Contains(inner)).ToList();
var innerNegativeRows = innerAttribute.NegativeRowNumbers
.Where(inner => attributeRows.Contains(inner)).ToList();
var positiveEntropy = innerPositiveRows.Count != 0
? MathExtension.GetBinaryLogaritm(innerPositiveRows.Count,
innerNegativeRows.Count + innerPositiveRows.Count)
: 0d;
var negativeEntropy = innerNegativeRows.Count != 0
? MathExtension.GetBinaryLogaritm(innerNegativeRows.Count,
innerNegativeRows.Count + innerPositiveRows.Count)
: 0d;
decisionColumn.CaseEntropy +=
(double)(innerNegativeRows.Count + innerPositiveRows.Count) / attributeRows.Count *
(-positiveEntropy - negativeEntropy);
}
}
else
{
decisionColumn.CaseEntropy = -1;
}
locallyUsedColumnIndexes.Add(Convert.ToInt32(decisionColumn.ColumnIndex));
}
Console.WriteLine("Nazwa atrybutu" + attribute.Value);
localAttributeValue = attribute.Value;
if (tempList.Any())
{
var x = tempList.MinBy(column => column.CaseEntropy);
if (attribute.PositiveRowNumbers.Count == 0 && attribute.NegativeRowNumbers.Count > 0)
{
Console.WriteLine("NOPE NOPE NOPE");
}
else if (x.CaseEntropy > -1)
{
Console.WriteLine("Index kolumny >>>> " + x.ColumnIndex);
selectedIndexesForNextRound.Add(Convert.ToInt32(x.ColumnIndex));
}
else
{
if (attribute.PositiveRowNumbers.Count == 0 && attribute.NegativeRowNumbers.Count > 0)
{
Console.WriteLine("Nie");
}
if (attribute.PositiveRowNumbers.Count > 0 && attribute.NegativeRowNumbers.Count == 0)
{
Console.WriteLine("Tak");
}
}
var currentBranchHierarchy = new List <int>();
if (fullHierarchy?[index] != null)
{
currentBranchHierarchy.AddRange(fullHierarchy[index]);
}
else
{
currentBranchHierarchy.Add(bestOneIndexes.First());
}
if (localAttributeValue != "-1" && x.CaseEntropy > -1)
{
var upperColumn = DecisionColumns[bestOneIndexes[index]];
currentBranchHierarchy.Add(upperColumn.Attributes.FindIndex(atr => atr.Value == localAttributeValue));
currentBranchHierarchy.Add(selectedIndexesForNextRound[index]);
newHierarchy.Add(currentBranchHierarchy);
}
}
else
{
var newThing = GetRowsFromPath(fullHierarchy[index]);
var positiveRows = attribute.PositiveRowNumbers.Intersect(newThing.PositiveRowNumbers).ToList();
var negativeRows = attribute.NegativeRowNumbers.Intersect(newThing.NegativeRowNumbers).ToList();
var x = attribute.NegativeRowNumbers.Count;
var y = attribute.PositiveRowNumbers.Count;
if (positiveRows.Count == 0 && negativeRows.Count > 0)
{
Console.WriteLine("Nie!");
}
else if (positiveRows.Count > 0 && negativeRows.Count == 0)
{
Console.WriteLine("Tak!");
}
}
}
//Sprzatanie tutaj?
Console.WriteLine("----------------------");
}
AlreadyUsedIndexes.AddRange(locallyUsedColumnIndexes);
if (selectedIndexesForNextRound.Count > 1)
{
DoUnknownThing(selectedIndexesForNextRound, newHierarchy);
}
}