public static BinaryClassificationPrediction RunBinaryClassification <TRowModel>(this Translation <TRowModel> translation, IDb db) where TRowModel : class, new()
{
if (translation.Operation.MiningOp != DataMinner.Mining.Enums.MiningOp.BinaryClassification)
{
throw new Exception("Cannot run binary classification on translation not meant for binary classification.");
}
// Gather initial data from the DB. We need this to train our binary classification operation
var sqlResults = new List <TRowModel>();
translation.RunSQL(db, (model) =>
{
sqlResults.Add(model);
});
// Here we are doing a binary classification operation
var binaryClassifier = new BinaryClassification <TRowModel>(sqlResults);
// Build the model so we can begin to use and classify
binaryClassifier.BuildModel();
BinaryClassificationData sentiment = new BinaryClassificationData {
SentimentText = translation.Operation.Text
};
var result = binaryClassifier.Evaluate(sentiment);
return(result);
}
private static int Main(string[] args)
{
// Test if input arguments were supplied.
Parser.Default.ParseArguments <Options>(args)
.WithParsed <Options>(o =>
{
// set logger
LoggingLevelSwitch levelSwitch = new LoggingLevelSwitch
{
MinimumLevel = LogEventLevel.Debug
};
Logger log = new LoggerConfiguration()
.MinimumLevel.ControlledBy(levelSwitch)
.WriteTo.Console(restrictedToMinimumLevel: LogEventLevel.Debug)
.WriteTo.File("backend_component_model_log-.txt", outputTemplate: "{Timestamp:yyyy-MM-dd HH:mm:ss.fff zzz} [{Level:u3}] {Message:lj}{NewLine}{Exception}",
rollingInterval: RollingInterval.Day, restrictedToMinimumLevel: LogEventLevel.Information)
.CreateLogger();
log.ForContext <MachineLearning>();
// load environmental variables
// DB connection configuration
JObject o1 = JObject.Parse(File.ReadAllText(@"C:\Users\Administrator\Documents\GitHub\machine-fault-diagnosis\backend\components\model\model_app\model_app\config\DBconfig.json"));
Environment.SetEnvironmentVariable("DBURL", (string)o1["DBURL"]);
Environment.SetEnvironmentVariable("DBUSER", (string)o1["DBUSER"]);
Environment.SetEnvironmentVariable("DBPW", (string)o1["DBPW"]);
// Model training configuration
JObject o2 = JObject.Parse(File.ReadAllText(@"C:\Users\Administrator\Documents\GitHub\machine-fault-diagnosis\backend\components\model\model_app\model_app\config\Trainconfig.json"));
Environment.SetEnvironmentVariable("ModelFile", (string)o2["ModelFile"]);
Environment.SetEnvironmentVariable("DataFile", (string)o2["DataFile"]);
Environment.SetEnvironmentVariable("TrainedModelFile", (string)o2["TrainedModelFile"]);
Environment.SetEnvironmentVariable("TrainingLogFile", (string)o2["TrainingLogFile"]);
if (o.mode == "I" || o.mode == "Interactive")
{
// Opens the interactive C# app
}
else
{
// Run machine learning engine in the background
switch (o.ModelType)
{
case "ProbabilisticNetwork":
log.Information("Start Probabilistic Graphical Model Engine");
Console.WriteLine("Please choose a task (Train, Inference or All)");
string t = Console.ReadLine();
string estimateStructure1;
if (t != "Inference")
{
Console.WriteLine("Please choose whether to estimated the graph structure or not ( Yes or No)");
estimateStructure1 = Console.ReadLine();
}
else
{
estimateStructure1 = "No";
}
PGM BNEngine = new PGM();
int result = BNEngine.Run(task: t, estimateStructure: estimateStructure1);
break;
case "TimeSeriesAnalysis":
log.Information("Start Time-series Analysis Engine");
break;
case "BinaryClassification":
log.Information("Start Binary Classification Engine");
string[] algorithms1 = { "AveragedPerceptronTrainer",
"SdcaLogisticRegressionBinaryTrainer",
"SdcaNonCalibratedBinaryTrainer",
"SymbolicSgdLogisticRegressionBinaryTrainer",
"LbfgsLogisticRegressionBinaryTrainer",
"LightGbmBinaryTrainer",
"FastTreeBinaryTrainer",
"FastForestBinaryTrainer",
"GamBinaryTrainer",
"FieldAwareFactorizationMachineTrainer",
"PriorTrainer",
"LinearSvmTrainer" };
Console.WriteLine("Please choose an algorithm from below");
Console.WriteLine(algorithms1);
string al1 = Console.ReadLine();
BinaryClassification bc = new BinaryClassification(al1);
int bcResult = bc.Run();
break;
case "MultiClassification":
log.Information("Start Multi Classification Engine");
string[] algorithms2 = { "LightGbmMulticlassTrainer",
"SdcaMaximumEntropyMulticlassTrainer",
"SdcaNonCalibratedMulticlassTrainer",
"LbfgsMaximumEntropyMulticlassTrainer",
"NaiveBayesMulticlassTrainer",
"OneVersusAllTrainer",
"PairwiseCouplingTrainer",
"ImageClassificationTrainer" };
Console.WriteLine("Please choose an algorithm from below");
Console.WriteLine(algorithms2);
string al2 = Console.ReadLine();
MultiClassification mc = new MultiClassification(al2);
int mcResult = mc.Run();
break;
case "Regression":
log.Information("Start Regression Engine");
string[] algorithms3 = { "LbfgsPoissonRegressionTrainer",
"LightGbmRegressionTrainer",
"SdcaRegressionTrainer",
"OlsTrainer",
"OnlineGradientDescentTrainer",
"FastTreeRegressionTrainer",
"FastTreeTweedieTrainer",
"FastForestRegressionTrainer",
"GamRegressionTrainer" };
Console.WriteLine("Please choose an algorithm from below");
Console.WriteLine(algorithms3);
string al3 = Console.ReadLine();
Regression rg = new Regression(al3);
int rgResult = rg.Run();
break;
case "Clustering":
log.Information("Start Clustering Engine. Use KMeans algorithm");
string al4 = "KMeans";
Clustering cluster = new Clustering(al4);
int clusResult = cluster.Run();
break;
case "AnomalyDetection":
log.Information("Start Anomaly Detection Engine. Use Randomized PCA algorithm");
string al5 = "RandomPCA";
Anomaly ano = new Anomaly(al5);
int anoResult = ano.Run();
break;
case "Ranking":
log.Information("Start Ranking Engine");
string[] algorithms6 = { "LightGbmRankingTrainer",
"FastTreeRankingTrainer" };
Console.WriteLine("Please choose an algorithm from below");
Console.WriteLine(algorithms6);
string al6 = Console.ReadLine();
Ranking rank = new Ranking(al6);
int rankResult = rank.Run();
break;
case "Recommendation":
log.Information("Start Recommendation Engine");
string al7 = "MatrixFactorizationTrainer";
Recommendation recomm = new Recommendation(al7);
int recommResult = recomm.Run();
break;
case "Forecast":
log.Information("Start Forecasting Engine");
string al8 = "ForecastBySsa";
Forecast forecast = new Forecast(al8);
int forecastResult = forecast.Run();
break;
default:
log.Information("Start AutoML Engine");
break;
}
}
}
).WithNotParsed(HandleParseError);
return(0);
}
static async Task BinaryClassifier(bool train = true)
{
var sqlConnection = $@"Server = localhost;database = Local;user = sa;password = sa";
var traindata = await Utilities.SQL.SQLServer.ExecuteReaderAsync <HeartData>(sqlConnection, $@"SELECT * FROM HeartTraining");
var testdata = await Utilities.SQL.SQLServer.ExecuteReaderAsync <HeartData>(sqlConnection, $@"SELECT * FROM HeartTest");
double accuracy = double.MinValue;
string bestAlg = string.Empty;
var mlContext = new MLContext();
var algorithms = new Dictionary <string, Func <IEnumerable <HeartData>, Action <ITransformer>, PredictionEngine <HeartData, HeartPredict> > >()
{
{ "FastTree", (data, action) => BinaryClassification.FastTree <HeartData, HeartPredict>(data, additionModelAction: action) },
{ "FastForest", (data, action) => BinaryClassification.FastForest <HeartData, HeartPredict>(data, additionModelAction: action) },
{ "SdcaLogisticRegression", (data, action) => BinaryClassification.SdcaLogisticRegression <HeartData, HeartPredict>(data, additionModelAction: action) },
};
foreach (var algorithm in algorithms)
{
PredictionEngine <HeartData, HeartPredict> engine = default;
ITransformer model = default;
var path = $@"BClassification_{algorithm.Key}.zip";
if (File.Exists(path) && !train)
{
model = MachineLearning.Global.LoadModel(path);
engine = mlContext.Model.CreatePredictionEngine <HeartData, HeartPredict>(model);
}
else
{
engine = algorithm.Value(traindata, (mdl) =>
{
model = mdl;
});
}
MachineLearning.Global.SaveModel(model, $@"Binaryclass_{algorithm.Key}.zip");
MachineLearning.ConsoleHelper.ConsoleWriteHeader($@"Evaluate metrics for {algorithm.Key} algorithm.");
try
{
var metrics = Metrics.EvaluateBinaryClassificationMetrics(model, mlContext.Data.LoadFromEnumerable(testdata), labelColumnName: nameof(HeartData.Label));
foreach (var prop in metrics.GetType().GetProperties())
{
Console.WriteLine($@"{prop.Name} : {prop.GetValue(metrics)}");
}
if (metrics.Accuracy > accuracy)
{
accuracy = metrics.Accuracy;
bestAlg = algorithm.Key;
}
}
catch (Exception e)
{
Console.WriteLine($"Unable to evaluate metrics : {e.Message}");
}
foreach (var t in testdata)
{
var predict = engine.Predict(t);
Console.WriteLine(string.Format(@"Actual {0,5} / Predict {1,5} with prob of {2,5}", t.Label, predict.Prediction, predict.Probability));
}
}
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine($@"Best algorithm based-on accuracy : {bestAlg}");
Console.ForegroundColor = ConsoleColor.White;
}
