public void LogisticRegressionTest()
{
var obsverationFile = @"\\protoapps\UserData\Jungkap\Joshua\score_training\observations.txt";
var categoriesFile = @"\\protoapps\UserData\Jungkap\Joshua\score_training\category_label.txt";
var obs = new List <double[]>();
var cat = new List <int>();
var line = "";
var logistic = new LogisticRegression();
System.IO.StreamReader file = new System.IO.StreamReader(obsverationFile);
while ((line = file.ReadLine()) != null)
{
var lineElements = line.Split('\t');
var doubleElements = new double[lineElements.Length];
for (var i = 0; i < lineElements.Length; i++)
{
doubleElements[i] = Convert.ToDouble(lineElements[i]);
}
obs.Add(doubleElements);
}
file = new System.IO.StreamReader(categoriesFile);
while ((line = file.ReadLine()) != null)
{
cat.Add(Convert.ToInt32(line));
}
var obsMatrix = obs.ToArray();
var classVector = cat.ToArray();
var weights = new[]
{
-3.94700432551079,
-0.304056338862474,
-0.908148491499137,
0.232858075077728,
0.340450694800618,
0.949161406416399,
0.268541376564235,
-0.0198709768288523,
-0.842410987704276,
0.196949733855582,
0.398403018300038,
0.860192152276247,
-0.0287180268260492,
-0.0178256644915106
};
var stopwatch = Stopwatch.StartNew();
weights = logistic.ComputeWeights(obsMatrix, classVector, weights);
stopwatch.Stop();
Console.WriteLine();
Console.WriteLine("Gradient Descent Time: {0}", stopwatch.ElapsedMilliseconds / 1000.0);
foreach (var x in weights)
{
Console.WriteLine(x);
}
Console.WriteLine();
var oneCount = 0;
var zeroCount = 0;
var totalRight = 0;
for (var i = 0; i < 10000; i++)
{
var score = CalculateClassificationScore(obsMatrix[i], weights);
var binary = GetClassification(score);
if (binary == cat[i])
{
totalRight++;
}
if (binary == 1)
{
oneCount++;
}
if (binary == 0)
{
zeroCount++;
}
}
Console.WriteLine("Scanning for 0");
Console.WriteLine("The number of 1: {0}", oneCount);
Console.WriteLine("The number of 0: {0}", zeroCount);
oneCount = 0;
zeroCount = 0;
for (var i = 10000; i < 20000; i++)
{
var score = CalculateClassificationScore(obsMatrix[i], weights);
var binary = GetClassification(score);
if (binary == cat[i])
{
totalRight++;
}
if (binary == 1)
{
oneCount++;
}
if (binary == 0)
{
zeroCount++;
}
}
Console.WriteLine("Scanning for 1");
Console.WriteLine("The number of 1: {0}", oneCount);
Console.WriteLine("The number of 0: {0}", zeroCount);
Console.WriteLine("Total Right: {0}", totalRight);
}