public TrainingResultRecord(int _id, ClassifierEnum _classifier, string _bestParamters, float _bestScore, float _accuracy, float _precision, float _recall, float _f1_score)
{
Id = _id;
Classifier = _classifier;
BestParamters = _bestParamters;
BestScore = _bestScore;
Accuracy = _accuracy;
Precision = _precision;
Recall = _recall;
F1_Score = _f1_score;
}
public void BindResultToView(string output)
{
var results = new Dictionary <ClassifierEnum, TrainingResultRecord>();
var lines = output.Split('\n');
var bestParamsSignal = "best parameters: ";
var bestScoreSignal = "best score: ";
var accuracySignal = "accuracy";
var weightedAvgSignal = "weighted avg";
ClassifierEnum currentClassifier = ClassifierEnum.COUNT;
for (int iLine = 0; iLine < lines.Length; iLine++)
{
var line = lines[iLine];
if (line.Contains("full training") || line.Contains("test model"))
{
for (int iClassifier = 0; iClassifier < (int)ClassifierEnum.COUNT; iClassifier++)
{
if (line.Contains(((ClassifierEnum)iClassifier).ToString()))
{
currentClassifier = (ClassifierEnum)iClassifier;
if (!results.ContainsKey(currentClassifier))
{
var nextId = results.Count;
results[currentClassifier] = new TrainingResultRecord();
results[currentClassifier].Id = nextId;
results[currentClassifier].Classifier = (ClassifierEnum)iClassifier;
}
break;
}
}
}
else if (line.Contains(bestParamsSignal))
{
results[currentClassifier].BestParamters = line.Substring(line.IndexOf(bestParamsSignal) + bestParamsSignal.Length).Trim();
}
else if (line.Contains(bestScoreSignal))
{
float.TryParse(line.Substring(line.IndexOf(bestScoreSignal) + bestScoreSignal.Length), out results[currentClassifier].BestScore);
}
else if (line.Contains(accuracySignal))
{
var parts = line.Substring(line.IndexOf(accuracySignal) + accuracySignal.Length).Split(' ')
.Select(x => x.Trim())
.Where(x => !string.IsNullOrWhiteSpace(x))
.ToArray();
if (parts.Length >= 1)
{
float.TryParse(parts[0], out results[currentClassifier].Accuracy);
}
}
else if (line.Contains(weightedAvgSignal))
{
var parts = line.Substring(line.IndexOf(weightedAvgSignal) + weightedAvgSignal.Length).Split(' ')
.Select(x => x.Trim())
.Where(x => !string.IsNullOrWhiteSpace(x))
.ToArray();
if (parts.Length >= 3)
{
float.TryParse(parts[0], out results[currentClassifier].Precision);
float.TryParse(parts[1], out results[currentClassifier].Recall);
float.TryParse(parts[2], out results[currentClassifier].F1_Score);
}
}
}
var resultDataGridView = _userControl.resultDataGridView;
resultDataGridView.Invoke((MethodInvoker) delegate
{
var table = new DataTable();
table.Columns.Add("Id", typeof(int));
table.Columns.Add("Classifier", typeof(ClassifierEnum));
table.Columns.Add("BestParamters", typeof(string));
table.Columns.Add("BestScore", typeof(float));
table.Columns.Add("Accuracy", typeof(float));
table.Columns.Add("Precision", typeof(float));
table.Columns.Add("Recall", typeof(float));
table.Columns.Add("F1_Score", typeof(float));
foreach (var result in results)
{
var dataRow = table.NewRow();
dataRow["Id"] = result.Value.Id;
dataRow["Classifier"] = result.Value.Classifier;
dataRow["BestParamters"] = result.Value.BestParamters;
dataRow["BestScore"] = result.Value.BestScore;
dataRow["Accuracy"] = result.Value.Accuracy;
dataRow["Precision"] = result.Value.Precision;
dataRow["Recall"] = result.Value.Recall;
dataRow["F1_Score"] = result.Value.F1_Score;
table.Rows.Add(dataRow);
}
resultDataGridView.DataSource = new BindingSource(table, null);
});
}