public static void CreateModel()
{
var context = new SiteKickContext();
var totalUniqueTechnology = (ulong)context.Technology.Select(x => x.Id).Count();
Console.WriteLine("Creating model...");
// load data from database
Console.WriteLine("Loading data...");
IDataView trainingDataView = mlContext.Data.LoadFromTextFile(
GetFullPath(CommonConstant.TRAINING_FILE_NAME),
columns: new[] {
new TextLoader.Column("Label", DataKind.Single, 0),
new TextLoader.Column(nameof(TechEntry.TechnologyId), DataKind.UInt32, keyCount: new KeyCount(CommonConstant.KEYCOUNT),
source: new [] { new TextLoader.Range(0) }),
new TextLoader.Column(nameof(TechEntry.CombinedTechnologyId), DataKind.UInt32, keyCount: new KeyCount(CommonConstant.KEYCOUNT),
source: new [] { new TextLoader.Range(1) }),
},
separatorChar: ',',
hasHeader: true);
var partitions = SplitData(trainingDataView, 0.2);
var pipeline = BuildTrainingPipeline();
var model = TrainModel(pipeline, partitions);
SaveModel(model, trainingDataView.Schema, GetFullPath(CommonConstant.MODEL_FILE_NAME));
EvaluateModel(model, partitions);
Console.WriteLine("Model created!");
}
public static void CreateTrainingData()
{
var entryList = new List <TechEntry>();
var context = new SiteKickContext();
var site = context.Site.Include(s => s.SiteTech).ThenInclude(st => st.Technology).ToList();
site.ForEach(s =>
{
var techList = s.SiteTech.Select(st => st.Technology).OrderBy(x => x.TechnologyGroupId).ToList();
var groupList = techList.Select(x => x.TechnologyGroupId).Distinct().ToList();
groupList.ForEach(g =>
{
var allTechInGroup = techList.Where(x => x.TechnologyGroupId == g).Select(x => x.Id).ToList();
allTechInGroup.ForEach(t =>
{
var techInGroupNotCurrent = allTechInGroup.Where(y => y != t).ToList();
techInGroupNotCurrent.ForEach(z =>
{
entryList.Add(new TechEntry
{
TechnologyId = (uint)t,
CombinedTechnologyId = (uint)z
});
});
});
});
});
CreateCSVFromList(entryList);
}
/// <summary>
/// Find the top 5 combined technology for the technology id
/// </summary>
public List <TechnologySuggestion> Predict(uint techId, int maxResult = 5)
{
_predictionEngine = _mlContext.Model.CreatePredictionEngine <TechEntry, TechnologySuggestion>(Model);
Console.WriteLine($"Calculating the top 5 technology for technology {techId} ...");
var context = new SiteKickContext();
var lastTechId = context.Technology.Select(x => x.Id).ToList().LastOrDefault();
var topResult = (from m in Enumerable.Range(1, lastTechId)
let predictions = _predictionEngine.Predict(new TechEntry
{
TechnologyId = techId,
CombinedTechnologyId = (uint)m
})
orderby predictions.Score descending
select(SuggestTechnologyId: m, Score: predictions.Score)).Select(x => new TechnologySuggestion()
{
TechnologyId = (uint)techId,
TechnologyName = context.Technology.Find((int)techId).Name,
SuggestTechnologyId = (uint)x.SuggestTechnologyId,
SuggestTechnologyName = context.Technology.Find(x.SuggestTechnologyId).Name,
Score = x.Score
}).Where(t => t.SuggestTechnologyId != techId).Take(maxResult).ToList();
return(topResult);
}
