MlContext doesn't understand the type of my temperature

Question

I'm Beginner at ML.NET and I have a little problem with my data. When I put them inside mlContext.Fit(...); Here's the error I received:

Column 'Temperature' has values of I4which is not the same as earlier observed type of R4.

Here's my code:

try
            {
                var mlContext = new MLContext();
                var reader = mlContext.Data.CreateTextReader<TrainData>(separatorChar: ',', hasHeader: false);
                var trainData = _context.Datas.Last();
                IDataView trainingdataView = reader.Read(Path.Combine(hostingEnvironment.WebRootPath, "data010220192341.txt"));
                var pipeline = mlContext.Transforms.Conversion.MapValueToKey("Delay")
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("StationDepart"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("StationArrival"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("Day"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("Train"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("WeatherText"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("HasPrecipitation"))
                    .Append(mlContext.Transforms.Categorical.OneHotEncoding("PrecipitationType"))
                    .Append(mlContext.Transforms.Concatenate("Features", "StationDepart", "StationArrival", "Day", "Train", "WeatherText", "Temperature", "Humidity", "HasPrecipitation", "PrecipitationType", "Time"))
                    .Append(mlContext.MulticlassClassification.Trainers.StochasticDualCoordinateAscent(labelColumn: "Delay", featureColumn: "Features"))
                    .Append(mlContext.Transforms.Conversion.MapKeyToValue("PredictedTime"));
                var model = pipeline.Fit(trainingdataView);
                var prediction = model.CreatePredictionEngine<TrainData, TrainPrediction>(mlContext).Predict(
                    new TrainData()
                    {
                        StationDepart = "Charleroi-Sud",
                        StationArrival = "Mons",
                        Day = "Friday",
                        Train = "BE.NMBS.IC3825",
                        WeatherText = "Partly cloudy",
                        Temperature = -1,
                        Humidity=0,
                        HasPrecipitation = false,
                        PrecipitationType=null,
                        Time=0444
                    });
                return prediction.PredictedTime.ToString();
            }
            catch (Exception e)
            {
                return e.Message;
            }

So when I get the Data on a text file, after that, I encode string Column and after I try to Train the model but it's at this moment I received the error. and my data are:

Charleroi-Sud,Mons,Thursday,BE.NMBS.IC3831,Partly sunny,-2,0,False,,1044,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC932,Mostly cloudy,-2,0,False,,1112,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC3832,Mostly cloudy,-1,0,False,,1144,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC933,Cloudy,-1,0,False,,1212,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC3842,Mostly cloudy,-1,0,False,,2144,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC943,Mostly cloudy,-1,0,False,,2212,0 Charleroi-Sud,Mons,Thursday,BE.NMBS.IC3843,Mostly cloudy,-1,0,False,,2247,0 Charleroi-Sud,Mons,Friday,BE.NMBS.IC3825,Partly cloudy,-1,0,False,,0444,0 Charleroi-Sud,Mons,Friday,BE.NMBS.IC3826,Mostly cloudy,-1,0,False,,0544,0 Charleroi-Sud,Mons,Friday,BE.NMBS.IC927,Mostly cloudy,-1,0,False,,0612,0

So as you can see, there is every ',' between data and the temperature is a int. And inside the TrainData, that done like that:

public class TrainData
    {
        [LoadColumn(0)]
        public string StationDepart { get; set; }
        [LoadColumn(1)]
        public string StationArrival { get; set; }
        [LoadColumn(2)]
        public string Day { get; set; }
        [LoadColumn(3)]
        public string Train { get; set; }
        [LoadColumn(4)]
        public string WeatherText { get; set; }
        [LoadColumn(5)]
        public int Temperature { get; set; }
        [LoadColumn(6)]
        public int Humidity { get; set; }
        [LoadColumn(7)]
        public bool HasPrecipitation { get; set; }
        [LoadColumn(8)]
        public string PrecipitationType { get; set; }
        [LoadColumn(9)]
        public int Time { get; set; }
        [LoadColumn(10)]
        public int Delay { get; set; }
    }

Maybe try changing the integers to floats? – Jon Feb 02 '19 at 03:56 — Jon, Feb 02 '19 at 03:56
Okay I will try that – thelittlewozniak Feb 02 '19 at 10:22 — thelittlewozniak, Feb 02 '19 at 10:22

score 1 · Accepted Answer · answered Feb 04 '19 at 13:59

The problem was that the Time, the delay and the temperature need to be in float and not in int.

public class TrainData
    {
        [LoadColumn(0)]
        public string StationDepart { get; set; }
        [LoadColumn(1)]
        public string StationArrival { get; set; }
        [LoadColumn(2)]
        public string Day { get; set; }
        [LoadColumn(3)]
        public string Train { get; set; }
        [LoadColumn(4)]
        public string WeatherText { get; set; }
        [LoadColumn(5)]
        public float Temperature { get; set; }
        [LoadColumn(6)]
        public float Humidity { get; set; }
        [LoadColumn(7)]
        public bool HasPrecipitation { get; set; }
        [LoadColumn(8)]
        public string PrecipitationType { get; set; }
        [LoadColumn(9)]
        public float Time { get; set; }
        [LoadColumn(10)]
        public float Delay { get; set; }
    }

MlContext doesn't understand the type of my temperature

1 Answers1