选择正确的算法或调整回归。FastTree获得预期结果

Question

我创建了一个ML项目。 我的数据集看起来像这样：

DateTime,DoorStatus,Metal_Temp,Burner_Input,Current_Roof_Temp,Roof_Temp_SetPo
23:35:00,   1,                   ,70.48548126,  678.4678345,    1142.275269
23:39:00,   1,                   ,99.99752045,  746.8652344,    1150
23:40:00,   1,                   ,100,          761.4857178,    1150
23:41:00,   1,          649      ,100,          768.6276855,    1150
23:42:00,   1,          650      ,100,          790.663208,     1150
23:43:00,   1,          651      ,100,          797.7381592,    1150
23:44:00,   1,          653      ,100,          817.0432129,    1150
23:45:00,   1,                   ,100,          847.1066895,    1150
23:46:00,   1,          656      ,100,          860.579834,     1150
23:47:00,   1,          657      ,100,          908.4719238,    1150
23:48:00,   1,          659      ,100,          936.2180176,    1150
23:49:00,   1,                   ,100,          945.9560547,    1150
23:50:00,   1,          663      ,100,          946.496582,     1150
23:51:00,   1,          664      ,100,          955.3989258,    1150
23:52:00,   1,          666      ,100,          974.0750732,    1150
23:53:00,   1,          668      ,100,          981.1708984,    1150
23:54:00,   1,          669      ,100,          996.7823486,    1150

该代码在编译和运行方面运行良好，但是它产生的预测结果与我的预期相去甚远！ 我试图通过输入时间，门状态，燃烧器输入和当前屋顶温度，设定点来预测Metal_Temp。预测输出Im得到例如。 0.012546、0.014568。

 static void Main(string[] args)
    {

        MLContext mLContext = new MLContext(seed: 0);        
        var model = Train(mLContext, @"C:\Projects\DotNetCoreProjects\MetalTempPredict\MetalTempPredict\Data\*");        
        int x = 1;
        int y = 0;
        while (x > y)
        {
            TestSinglePredicition(mLContext, model);
            x = x + 1;
            y = y + 1;
        };
    }

    public static ITransformer Train(MLContext mLContext, string dataPath)
    {
        IDataView dataView = mLContext.Data.LoadFromTextFile<LargData>(@"C:\Projects\DotNetCoreProjects\MetalTempPredict\MetalTempPredict\Data\*", hasHeader: true, separatorChar: ',');

        var pipeline = mLContext.Transforms.CopyColumns(outputColumnName: "Label", inputColumnName: "Metal_Temp")
            .Append(mLContext.Transforms.Categorical.OneHotHashEncoding(outputColumnName: "date_timeEnocded", inputColumnName: "date_time"))
            .Append(mLContext.Transforms.Categorical.OneHotEncoding(outputColumnName: "DoorStatusEncoded", inputColumnName: "DoorStatus"))                        
            .Append(mLContext.Transforms.Concatenate("Features", "DoorStatusEncoded", "date_timeEnocded", "Metal_Temp", "Burner_Input", "Current_Roof_Temp", "Roof_Temp_SetPoint"))
            .Append(mLContext.Regression.Trainers.FastTree());

        var model = pipeline.Fit(dataView);

        mLContext.Model.Save(model, dataView.Schema, @"C:\Projects\DotNetCoreProjects\MetalTempPredict\MetalTempPredict\Data\Files" + DateTime.Now.ToString("yyyy - MM - dd HH - mm - ss") + ".zip");
        return model;
    }

    private static void Evaluate(MLContext mLContext, ITransformer model)
    {
        IDataView dataView = mLContext.Data.LoadFromTextFile<LargData>(@"C:\Projects\DotNetCoreProjects\MetalTempPredict\MetalTempPredict\Data\TestData\*", hasHeader: true, separatorChar: ',');
        var predictions = model.Transform(dataView);
        var metrics = mLContext.Regression.Evaluate(predictions, "Label", "Score");

        Console.WriteLine();
        Console.WriteLine($"***********************************");
        Console.WriteLine($" Model quality metrics evaluation  ");
        Console.WriteLine($"*----------------------------------");
        Console.WriteLine($"* (RSquared takes values between 0 and 1. The closer its value is to 1, the better the model is.) RSquared Score: {metrics.RSquared:#.##}");
        Console.WriteLine($"  (RMS is one of the evaluation metrics of the regression model. The lower it is, the better the model is.) Root means Squared Error:  {metrics.RootMeanSquaredError:#.##}");
    }

    private static void TestSinglePredicition(MLContext mLContext, ITransformer model)
    {

        var predictionFunction = mLContext.Model.CreatePredictionEngine<LargData, GetMetalTemp>(model);

        var SamplePredict = new LargData();         

        Console.WriteLine("Time Input");
        SamplePredict.date_time = (Console.ReadLine()).ToString();

        Console.WriteLine("Door Status");
        SamplePredict.DoorStatus = int.Parse(Console.ReadLine());

        Console.WriteLine("Enter burner input");
        SamplePredict.Burner_Input = float.Parse(Console.ReadLine());

        Console.WriteLine("Enter Curr roof temp");
        SamplePredict.Current_Roof_Temp = float.Parse(Console.ReadLine());

        Console.WriteLine("Enter Roof Temp set point");
        SamplePredict.Roof_Temp_SetPoint = float.Parse(Console.ReadLine());            

        var prediction = predictionFunction.Predict(SamplePredict);
        Console.WriteLine($"******************************");
        Console.WriteLine($"Predicted Metal_Temp: {prediction.Metal_Temp}");
        Console.WriteLine($"******************************");
        Console.ReadLine();    
    }        
}
}