public Accord.MachineLearning.DecisionTrees.DecisionTree Learn(string[][] records, string[] columnNamesWithoutResult, ref Codification codebook)
{
DataTable data = new DataTable();
foreach (var columnName in records[0])
{
data.Columns.Add(columnName);
}
int rowsAdderCounter = 0;
foreach (var record in records)
{
if (rowsAdderCounter == 0)
{
rowsAdderCounter++;
continue;
}
data.Rows.Add(record);
}
double[][] inputs = data.ToJagged(columnNamesWithoutResult);
string[] labels = data.ToArray <string>(Constants.RESULT_COLUMN_NAME);
int[] outputs = codebook.Translate(Constants.RESULT_COLUMN_NAME, labels);
var teacher = new C45Learning();
foreach (var columnName in columnNamesWithoutResult)
{
DecisionVariable decVar = new DecisionVariable(columnName, DecisionVariableKind.Continuous);
teacher.Add(decVar);
}
Accord.MachineLearning.DecisionTrees.DecisionTree tree = teacher.Learn(inputs, outputs);
return(tree);
}
public ClassifierTitanic()
{
rawData = new DataTable("Titanic Data");
trainingData = new DataTable();
testingData = new DataTable();
string filedata = System.IO.File.ReadAllText("../titanicData.txt");
string[] dataColumns = System.IO.File.ReadAllText("../titanicColumns.txt").Split(',');
//Input columns are to be learned from
string[] inputColumns = new string[dataColumns.Length - 1];
Array.Copy(dataColumns, 0, inputColumns, 0, dataColumns.Length - 1);
//Output is what we are trying to predict
string outputColumn = dataColumns[dataColumns.Length - 1];
//Create an easy way to store and manipulate data
rawData.Columns.Add(inputColumns);
rawData.Columns.Add(outputColumn);
trainingData.Columns.Add(inputColumns);
trainingData.Columns.Add(outputColumn);
testingData.Columns.Add(inputColumns);
testingData.Columns.Add(outputColumn);
string[] lines = filedata.Split(
new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (var line in lines)
{
rawData.Rows.Add(line.Split(','));
}
//Clean up data representation and missing data
rawData = cleanData(rawData);
DataTable[] dt = splitDataForTraining(rawData, .8, inputColumns, outputColumn);
trainingData = dt[0];
testingData = dt[1];
//---------
codebook = new Codification(trainingData);
DataTable symbols = codebook.Apply(trainingData);
int[][] inputs = symbols.ToJagged <int>("Pclass", "Title", "Sex", "Age", "SibSp", "Parch", "Fare", "Cabin", "Embarked");
int[] outputs = symbols.ToArray <int>("Survived");
// We can either specify the decision attributes we want
// manually, or we can ask the codebook to do it for us:
DecisionVariable[] attributes = DecisionVariable.FromCodebook(codebook, inputColumns);
// Create a teaching algorithm:
var teacher = new C45Learning();
teacher.Add(attributes[0]);
teacher.Add(attributes[1]);
teacher.Add(attributes[4]);
teacher.Add(new DecisionVariable("Age", new DoubleRange(0, 99)));
teacher.Add(new DecisionVariable("SibSp", new DoubleRange(0, 10)));
teacher.Add(new DecisionVariable("Parch", new DoubleRange(0, 10)));
teacher.Add(new DecisionVariable("Fare", new DoubleRange(0, 400)));
teacher.Add(attributes[10]);
teacher.Add(attributes[11]);
// and induce a decision tree from the data:
DecisionTree tree = teacher.Learn(inputs, outputs);
// To get the estimated class labels, we can use
int[] predicted = tree.Decide(inputs);
// Moreover, we may decide to convert our tree to a set of rules:
DecisionSet rules = tree.ToRules();
// And using the codebook, we can inspect the tree reasoning:
string ruleText = rules.ToString(codebook, "Survived",
System.Globalization.CultureInfo.InvariantCulture);
foreach (DataRow d in testingData.Rows)
{
int[] tempVars = codebook.Transform(new[, ]
{
{ "Pclass", d[0].ToString() },
{ "Title", d[1].ToString() },
{ "Sex", d[4].ToString() },
{ "Cabin", d[10].ToString() },
{ "Embarked", d[11].ToString() }
});
int[] query =
{
tempVars[0],
tempVars[1],
tempVars[2],
int.Parse(d[5].ToString()),
int.Parse(d[6].ToString()),
int.Parse(d[7].ToString()),
int.Parse(d[9].ToString()),
tempVars[3],
tempVars[4]
};
int predictedValue = tree.Decide(query);
int actualValue = int.Parse(d[12].ToString());
if (predictedValue == actualValue)
{
if (actualValue == 1)
{
truePositives++;
}
else
{
trueNegatives++;
}
}
else
{
if (actualValue == 1)
{
falseNegatives++;
}
else
{
falsePositives++;
}
}
}
var dasdfasd = 5;
//// And the classification error (of 0.0) can be computed as
//double error = new ZeroOneLoss(outputs).Loss(tree.Decide(inputs));
//// To compute a decision for one of the input points,
//// such as the 25-th example in the set, we can use
////
//int y = tree.Decide(inputs[25]); // should be 1
//int[][] inputs = symbols.ToJagged<int>("???");
//int[] outputs = symbols.ToArray<int>("Survived");
//string[] decisionVariables = { "???" };
//DecisionVariable[] attributes = DecisionVariable.FromCodebook(codebook, decisionVariables);
//// Create a teacher ID3 algorithm
//var id3learning = new ID3Learning(attributes);
//tree = id3learning.Learn(inputs, outputs);
//// Compute the training error when predicting training instances
//double error = new ZeroOneLoss(outputs).Loss(tree.Decide(inputs));
}
