private static async Task Verify()
{
KarmaPredictor predictor = new KarmaPredictor();
if (!predictor.IsModelTrained())
{
return;
}
predictor.LoadModel();
Func <string, string[]> filter = w => Regex.Replace(w, @"[^a-zA-Z]", " ").Trim().ToLowerInvariant().Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries).Where(d => d.Count() < 20).ToArray();
var client = new MongoClient("mongodb://localhost:27017");
var db = client.GetDatabase("active_app_record");
var trainingCollection = db.GetCollection <AppUsageRecord>("daily_records");
var cursor = await trainingCollection.FindAsync(Builders <AppUsageRecord> .Filter.Eq(f => f.Id, AppUsageRecord.GetGeneratedId(DateTime.Now)));
var trainingData = await cursor.FirstOrDefaultAsync();
var processInfos = trainingData.ActiveApps.Select(f => f.Value).ToArray();
var predictions = predictor.Predict(processInfos);
for (int i = 0; i < processInfos.Length; i++)
{
Console.WriteLine($"for window title [{processInfos[i].MainWindowTitle}] result: {predictions[i]}");
}
}
private static async Task Train()
{
KarmaPredictor predictor = new KarmaPredictor();
if (predictor.IsModelTrained())
{
return;
}
Func <string, string[]> filter = w => Regex.Replace(w, @"[^a-zA-Z]", " ").Trim().ToLowerInvariant().Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries).Where(d => d.Count() < 20).ToArray();
var client = new MongoClient("mongodb://localhost:27017");
var db = client.GetDatabase("active_app_record");
var trainingCollection = db.GetCollection <ProcessInfoLabeledItem>("training_data");
var cursor = await trainingCollection.FindAsync(Builders <ProcessInfoLabeledItem> .Filter.Empty);
var trainingData = await cursor.ToListAsync();
predictor.TrainModel(trainingData);
predictor.SaveModel();
Console.WriteLine("Learning completed, save trained model to disk");
}
