private void Train(HashSet <string> options)
{
if (options == null)
{
options = new HashSet <string>(new string [] { "b" });
}
// train with bayes model
if (options.Contains("b"))
{
var trainer = new BayesModel((string)GlobalParameter.Get(DefaultParameter.Field.train_feature_file),
(string)GlobalParameter.Get(DefaultParameter.Field.model_file));
try
{
trainer.Train();
}
catch (Exception e)
{
Console.WriteLine("Error occur during train for " + e.Message);
throw new Exception();
}
}
}
public void Test()
{
if (model == null)
{
Initial();
}
var fields = BayesModel.GetFields(sourceFile);
FileReader reader = new LargeFileReader(sourceFile);
FileWriter writer = new LargeFileWriter(resultFile, FileMode.Create);
// actual label-->(prediced label-->times)
var detailDic = new Dictionary <string, Dictionary <string, int> >();
var positiveNums = new Dictionary <string, int>(); // positive number by type
var predictedNums = new Dictionary <string, int>(); // predicted number by type
var actualNums = new Dictionary <string, int>(); // actual number by type
Dictionary <string, int> dic = null;
Pair <string, Dictionary <string, object> > feature = null;
var i = 0;
while ((feature = BayesModel.GetFeatureItem(reader, fields)) != null)
{
i++;
var label = feature.first;
string predictedLabel = null;
try
{
predictedLabel = Predict(feature.second);
}
catch (Exception)
{
Console.WriteLine("Wrong!");
writer.WriteLine(i + "\t" + label + "\tNULL");
continue;
}
writer.Write(string.Format("{0}\t{1, -30}", i, label));
foreach (var score in this.scores)
{
writer.Write(string.Format("{0,30}:{1,-10:F2}", score.first, score.second));
}
writer.Write("\r");
if (label.Equals(predictedLabel))
{
try
{
positiveNums[label] += 1;
}
catch (Exception)
{
positiveNums[label] = 1;
}
}
try
{ // update predicted number
predictedNums[predictedLabel] += 1;
}
catch (Exception)
{
predictedNums[predictedLabel] = 1;
}
try
{ // update actually number
actualNums[label] += 1;
}
catch (Exception)
{
actualNums[label] = 1;
}
// update detail dictionary
try
{
dic = detailDic[label];
}
catch (Exception)
{
dic = new Dictionary <string, int>();
detailDic[label] = dic;
}
try
{
dic[predictedLabel] += 1;
}
catch (Exception)
{
dic[predictedLabel] = 1;
}
}
var buffer = new StringBuilder();
buffer.Append(string.Format("{0,-30}", "actual label |predicted type"));
foreach (var key in this.labels)
{
buffer.Append(string.Format("{0,-30}", key));
}
buffer.Append(string.Format("{0,-30}\r", "recall"));
foreach (var key in this.labels)
{
buffer.Append(string.Format("{0,-30}", key));
dic = detailDic[key];
foreach (var k in this.labels)
{
buffer.Append(string.Format("{0,-30}", dic[k]));
}
// recall
buffer.Append(string.Format("{0,-30}\r", 1.0 * positiveNums[key] / actualNums[key]));
}
buffer.Append(string.Format("{0,-30}", "precision"));
foreach (var key in this.labels)
{
buffer.Append(string.Format("{0,-30:f5}", 1.0 * positiveNums[key] / predictedNums[key]));
}
buffer.Append("\r");
writer.WriteLine(buffer.ToString());
writer.Close();
}
