// (0.5,0.5):
// weight distribution = Gaussian(-0.02787, 0.2454)
// error rate = 0.0527452589744697 = 1221/23149
// (1,1):
// weight distribution = Gaussian(-0.03117, 0.3967)
// error rate = 0.0522268780508877 = 1209/23149
// (2,2):
// weight distribution = Gaussian(-0.03522, 0.6794)
// error rate = 0.0530476478465593 = 1228/23149
// (10,10):
// weight distribution = Gaussian(-0.05455, 2.96)
// error rate = 0.0580586634411854 = 1344/23149
#if SUPPRESS_UNREACHABLE_CODE_WARNINGS
#pragma warning restore 162
#endif
public static void Rcv1Test2()
{
GaussianArray wPost;
Gaussian biasPost;
BinaryFormatter serializer = new BinaryFormatter();
//TODO: change path
using (Stream stream = File.OpenRead(@"c:\Users\minka\Downloads\rcv1\weights.bin"))
{
wPost = (GaussianArray)serializer.Deserialize(stream);
biasPost = (Gaussian)serializer.Deserialize(stream);
}
if (true)
{
GaussianEstimator est = new GaussianEstimator();
foreach (Gaussian item in wPost)
{
est.Add(item.GetMean());
}
Console.WriteLine("weight distribution = {0}", est.GetDistribution(new Gaussian()));
}
var predict = new BpmPredict2();
predict.SetPriors(wPost, biasPost);
int count = 0;
int errors = 0;
//TODO: change path
foreach (Instance instance in new VwReader(@"c:\Users\minka\Downloads\rcv1\rcv1.test.vw.gz"))
{
bool yPred = predict.Predict(instance);
if (yPred != instance.label)
{
errors++;
}
count++;
}
Console.WriteLine("error rate = {0} = {1}/{2}", (double)errors / count, errors, count);
}
public static void Rcv1Test(double wVariance, double biasVariance)
{
int count = 0;
if (false)
{
int maxFeatureIndex = 0;
foreach (Instance instance in new VwReader(Path.Combine(dataFolder, "rcv1.train.vw.gz")))
{
count++;
if (count % 10000 == 0)
{
Console.WriteLine(count);
}
foreach (int index in instance.featureIndices)
{
if (index > maxFeatureIndex)
{
maxFeatureIndex = index;
}
}
}
Console.WriteLine("{0} features", maxFeatureIndex + 1);
}
int nf = 47152;
var train = new BpmTrain2();
var predict = new BpmPredict2();
train.SetPriors(nf, wVariance, biasVariance);
StreamWriter writer = new StreamWriter(Path.Combine(dataFolder, "log.txt"));
int errors = 0;
//int errors2 = 0;
//StreamReader reader = new StreamReader(Path.Combine(dataFolder, "preds.txt");
// takes 92s to train
// takes 74s just to read the data
// takes 15s just to do 'wc' on the data
// there are 781265 data points in train, 23149 in test
foreach (Instance instance in new VwReader(Path.Combine(dataFolder, "rcv1.train.vw.gz")))
{
predict.SetPriors(train.wPost, train.biasPost);
bool yPred = predict.Predict(instance);
if (yPred != instance.label)
{
errors++;
}
//double pred2 = double.Parse(reader.ReadLine());
//if ((pred2 > 0.5) != instance.label) errors2++;
train.Train(instance);
count++;
if (count % 1000 == 0)
{
Console.WriteLine("{0} {1} {2}", count, (double)errors / count, train.biasPost);
//Console.WriteLine("{0} {1} {2} {3}", count, (double)errors/count, (double)errors2/count, train.biasPost);
writer.WriteLine("{0} {1}", count, (double)errors / count);
writer.Flush();
//if (count == 10000) break;
}
}
writer.Dispose();
#if NETFRAMEWORK
// In the .NET 5.0 BinaryFormatter is obsolete
// and would produce errors. This test code should be migrated.
// See https://aka.ms/binaryformatter
if (true)
{
BinaryFormatter serializer = new BinaryFormatter();
using (Stream stream = File.Create(Path.Combine(dataFolder, "weights.bin")))
{
serializer.Serialize(stream, train.wPost);
serializer.Serialize(stream, train.biasPost);
}
}
#endif
}
#pragma warning disable 162
#endif
public static void Rcv1Test(double wVariance, double biasVariance)
{
int count = 0;
if (false)
{
int maxFeatureIndex = 0;
//TODO: change path
foreach (Instance instance in new VwReader(@"c:\Users\minka\Downloads\rcv1\rcv1.train.vw.gz"))
{
count++;
if (count % 10000 == 0)
{
Console.WriteLine(count);
}
foreach (int index in instance.featureIndices)
{
if (index > maxFeatureIndex)
{
maxFeatureIndex = index;
}
}
}
Console.WriteLine("{0} features", maxFeatureIndex + 1);
}
int nf = 47152;
var train = new BpmTrain2();
var predict = new BpmPredict2();
train.SetPriors(nf, wVariance, biasVariance);
//TODO: change path
StreamWriter writer = new StreamWriter(@"c:\Users\minka\Downloads\rcv1\log.txt");
int errors = 0;
//int errors2 = 0;
//TODO: change path
//StreamReader reader = new StreamReader(@"c:\Users\minka\Downloads\rcv1\preds.txt");
// takes 92s to train
// takes 74s just to read the data
// takes 15s just to do 'wc' on the data
// there are 781265 data points in train, 23149 in test
//TODO: change path
foreach (Instance instance in new VwReader(@"c:\Users\minka\Downloads\rcv1\rcv1.train.vw.gz"))
{
predict.SetPriors(train.wPost, train.biasPost);
bool yPred = predict.Predict(instance);
if (yPred != instance.label)
{
errors++;
}
//double pred2 = double.Parse(reader.ReadLine());
//if ((pred2 > 0.5) != instance.label) errors2++;
train.Train(instance);
count++;
if (count % 1000 == 0)
{
Console.WriteLine("{0} {1} {2}", count, (double)errors / count, train.biasPost);
//Console.WriteLine("{0} {1} {2} {3}", count, (double)errors/count, (double)errors2/count, train.biasPost);
writer.WriteLine("{0} {1}", count, (double)errors / count);
writer.Flush();
//if (count == 10000) break;
}
}
writer.Dispose();
if (true)
{
BinaryFormatter serializer = new BinaryFormatter();
//TODO: change path
using (Stream stream = File.Create(@"c:\Users\minka\Downloads\rcv1\weights.bin"))
{
serializer.Serialize(stream, train.wPost);
serializer.Serialize(stream, train.biasPost);
}
}
}