SkillDatingMatrixFactorization

interests_train.csv

SkillDatingMatrixFactorization.Program.cs

using Microsoft.ML;
using Microsoft.ML.Data;
using Microsoft.ML.Trainers;
using System;
using System.IO;

// In machine learning, the columns that are used to make a prediction are called Features,
// and the column with the returned prediction is called the Label.

// We want to predict interest ratings => Label
// (Name, Interest) => Features ==> Predict the label


// Load
var mlContext = new MLContext();
var trainingDataPath = Path.Combine(Environment.CurrentDirectory, "interests_train.csv");
var trainingDataView = mlContext.Data.LoadFromTextFile<StudentInterest>(trainingDataPath, hasHeader: true, separatorChar: ',');

// Build and train model
var estimator = mlContext.Transforms.Conversion.MapValueToKey("nameEncoded", "Name")
    .Append(mlContext.Transforms.Conversion.MapValueToKey("interestEncoded", "Interest"));

var options = new MatrixFactorizationTrainer.Options
{
    MatrixColumnIndexColumnName = "nameEncoded",
    MatrixRowIndexColumnName = "interestEncoded",
    LabelColumnName = "Label",
    NumberOfIterations = 20,
    ApproximationRank = 100
};

var trainerEstimator = estimator.Append(mlContext.Recommendation().Trainers.MatrixFactorization(options));

var model = trainerEstimator.Fit(trainingDataView);

// Predict
var predictionEngine = mlContext.Model.CreatePredictionEngine<StudentInterest, StudentInterestPrediction>(model);
var testInput = new StudentInterest { Name = "Gino", Interest = "Datenbanken" };
var prediction = predictionEngine.Predict(testInput);

Console.WriteLine(prediction.Score); // -> 2,2039702 just about right
Console.ReadLine();


public class StudentInterest
{
    [LoadColumn(0)]
    public string Name;

    [LoadColumn(1)]
    public string Interest;

    [LoadColumn(2)]
    public float Label;
}

public class StudentInterestPrediction
{
    public float Label;
    public float Score;
}