Myujth

There doesn't seem to be much documentation on ML.Net as it appears to be relatively new. I've been running into problem after problem trying to learn how to use it, and I finally figured out it enough to get it to at least run without experiencing an error; however, there seems to be a problem with my model. It always returns 0 with a probability of 50%. I've included my code below. Does anyone know any good resources for the most recent version of ML.Net that I can explore? The code below is supposed to be making a binary classification model that can predict whether a team is going or not going to the playoff. The data is just the final results of last season with most of the data removed so the only columns that remain are the average age, wins, losses, and playoff status (1 = playoffs & 0 = no playoff).

Program.cs

using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using Microsoft.ML;
using Microsoft.ML.Core.Data;
using Microsoft.ML.Runtime.Api;
using Microsoft.ML.Runtime.Data;

namespace MachineLearning2

 class Program
 
 static readonly string _trainDataPath = Path.Combine(Environment.CurrentDirectory, "trainingNHL.txt");
 static readonly string _testDataPath = Path.Combine(Environment.CurrentDirectory, "testingNHL.txt");
 static readonly string _modelPath = Path.Combine(Environment.CurrentDirectory, "Model.zip");
 static TextLoader _textLoader;

 static void Main(string args)
 
 MLContext mlContext = new MLContext(seed: 0);
 _textLoader = mlContext.Data.TextReader(new TextLoader.Arguments()
 
 Separator = ",",
 HasHeader = false,
 Column = new
 
 new TextLoader.Column("Age", DataKind.R4, 0),
 new TextLoader.Column("Wins", DataKind.R4, 1),
 new TextLoader.Column("Losses", DataKind.R4, 2),
 new TextLoader.Column("Label", DataKind.R4, 3)
 
 );
 var model = Train(mlContext, _trainDataPath);
 Evaluate(mlContext, model);
 Predict(mlContext, model);
 PredictWithModelLoadedFromFile(mlContext);

 

 public static ITransformer Train(MLContext mlContext, string dataPath)
 
 IDataView dataView = _textLoader.Read(dataPath);
 var pipeline = mlContext.Transforms.Concatenate("Features","Age", "Wins", "Losses")
 .Append(mlContext.BinaryClassification.Trainers.FastTree(numLeaves: 50, numTrees: 50, minDatapointsInLeafs: 20));
 Console.WriteLine("=============== Create and Train the Model ===============");
 var model = pipeline.Fit(dataView);
 Console.WriteLine("=============== End of training ===============");
 Console.WriteLine();
 return model;
 

 public static void Evaluate(MLContext mlContext, ITransformer model)
 
 IDataView dataView = _textLoader.Read(_testDataPath);
 Console.WriteLine("=============== Evaluating Model accuracy with Test data===============");
 var predictions = model.Transform(dataView);
 var metrics = mlContext.BinaryClassification.Evaluate(predictions, "Label");
 Console.WriteLine();
 Console.WriteLine("Model quality metrics evaluation");
 Console.WriteLine("--------------------------------");
 Console.WriteLine($"Accuracy: metrics.Accuracy:P2");
 Console.WriteLine($"Auc: metrics.Auc:P2");
 Console.WriteLine($"F1Score: metrics.F1Score:P2");
 Console.WriteLine("=============== End of model evaluation ===============");
 SaveModelAsFile(mlContext, model);
 

 private static void SaveModelAsFile(MLContext mlContext, ITransformer model)
 
 using (var fs = new FileStream(_modelPath, FileMode.Create, FileAccess.Write, FileShare.Write))
 mlContext.Model.Save(model, fs);
 Console.WriteLine("The model is saved to 0", _modelPath);

 

 public static void Predict(MLContext mlContext, ITransformer model)
 Prediction: (Convert.ToBoolean(resultprediction.Prediction) ? "Yes" : "No") 

 public static void PredictWithModelLoadedFromFile(MLContext mlContext)
 
 IEnumerable<NHLData> teams = new
 
 new NHLData
 
 Age = 29,
 Wins = 30,
 Losses = 52
 ,
 new NHLData
 
 Age = 35,
 Wins = 80,
 Losses = 2
 
 ;
 ITransformer loadedModel;
 using (var stream = new FileStream(_modelPath, FileMode.Open, FileAccess.Read, FileShare.Read))
 
 loadedModel = mlContext.Model.Load(stream);
 
 // Create prediction engine
 var nhlStreamingDataView = mlContext.CreateStreamingDataView(teams);
 var predictions = loadedModel.Transform(nhlStreamingDataView);

 // Use the model to predict whether comment data is toxic (1) or nice (0).
 var predictedResults = predictions.AsEnumerable<NHLPrediction>(mlContext, reuseRowObject: false);
 Console.WriteLine();

 Console.WriteLine("=============== Prediction Test of loaded model with a multiple samples ===============");
 var teamsAndPredictions = teams.Zip(predictedResults, (team, prediction) => (team, prediction));
 foreach (var item in teamsAndPredictions)
 Prediction: (Convert.ToBoolean(item.prediction.Prediction) ? "Yes" : "No") 
 Console.WriteLine("=============== End of predictions ===============");
 
 

NHLData.cs

using System;
using System.Collections.Generic;
using System.Text;
using Microsoft.ML.Runtime.Api;

namespace MachineLearning2

 public class NHLData
 
 [Column(ordinal: "0")]
 public float Age;
 [Column(ordinal: "1")]
 public float Wins;
 [Column(ordinal: "2")]
 public float Losses;
 [Column(ordinal: "3", name: "Label")]
 public float Playoffs;
 

 public class NHLPrediction
 
 [ColumnName("PredictedLabel")]
 public bool Prediction get; set; 

 [ColumnName("Probability")]
 public float Probability get; set; 

 [ColumnName("Score")]
 public float Score get; set; 
 

trainingNHL.txt (columns: Age, Wins, Losses, Playoffs)

28.4,53,18,1
27.5,54,23,1
28,51,24,1
28.3,49,26,1
29.5,45,26,1
28.8,45,27,1
29.1,45,29,1
27.7,44,29,1
26.4,43,30,1
28.5,42,32,0
27,36,35,0
26.8,36,40,0
28,33,39,0
30.2,30,39,0
26.5,29,41,0
27.1,25,45,0

testingNHL.txt (columns: Age, Wins, Losses, Playoffs)

26.8,52,20,1
28.6,50,20,1
28.4,49,26,1
28.7,44,25,1
27.7,47,29,1
27.4,42,26,1
26.4,45,30,1
27.8,44,30,0
28.5,44,32,0
28.4,37,35,0
28.4,35,37,0
28.7,34,39,0
28.2,31,40,0
27.8,29,40,0
29.3,28,43,0

Is trainingNHL.txt the full dataset you are using or only a sample of it? I just tried training on it with FastTree and I see "Warning: 50 of the boosting iterations failed to grow a tree. This is commonly because the minimum documents in leaf hyperparameter was set too high for this dataset."

Given the parameters you set in FastTree, you will need more data to train a meaningful model. When I change minDatapointsInLeafs to 2, I am able to train a non-trivial model (although the results are still not extremely trustworthy due to the amount of data). You could also try using something like AveragedPerceptron or SDCA.