Revize 662b2404

Re #8832 integration

Server/ServerApp/Predictor/FeatureExtractor.cs

@@ -52 +52 @@
             }
 
             var res =  new List<ModelInput>();
-            // TODO use iterators to access records about attendance and weather together
-            foreach (Parser.OutputInfo.ActivityInfo val in dataParser.AttendanceList) {
-                if (buildings.Contains(val.building)) {
+            // use enumerator to access records about attendance and weather together
+            IEnumerator<Parser.OutputInfo.WeatherInfo> weatherEnumerator = dataParser.WeatherList.GetEnumerator();
+            weatherEnumerator.MoveNext();
+            foreach (Parser.OutputInfo.ActivityInfo val in dataParser.AttendanceList)
+            {
+                if (!weatherEnumerator.Current.startTime.Equals(val.startTime))
+                {
+                    weatherEnumerator.MoveNext();
+                }
+                if (buildings.Contains(val.building))
+                {
+                    Parser.OutputInfo.WeatherInfo weatherInfo = weatherEnumerator.Current;
                     res.Add(new ModelInput
                     {
-                        Temp = 0,
+                        Temp = (float)weatherInfo.temp,
                         Time = val.startTime,
+                        Wind = (float)weatherInfo.wind,
+                        Rain = (float)weatherInfo.rain,
                         // TODO convert val.amount to label
                     });
                 }

Server/ServerApp/Predictor/IPredictionController.cs

@@ -17 +17 @@
     {
         /// <summary>
         /// </summary>
-        /// <returns>A dictionary with all existing predictors.</returns>
+        /// <returns>A list with all existing predictors.</returns>
         List<string> GetPredictors();
 
         /// <summary>

Server/ServerApp/Predictor/IPredictor.cs

@@ -30 +30 @@
         /// <param name="input">A collection of model feature vectors in <c>ModelInput</c> instances.</param>
         /// <returns>A collection of <c>PredictionResult</c> instances.</returns>
         IDataView Predict(IEnumerable<ModelInput> input);
+
+        // TODO define Save method
     }
 }

Server/ServerApp/Predictor/ModelInput.cs

@@ -30 +30 @@
         /// </summary>
         [ColumnName("Time"), LoadColumn(2)]
         public DateTime Time { get; set; }
+
+        /// <summary>
+        /// Wind velocity in ? units
+        /// </summary>
+        [ColumnName("Wind"), LoadColumn(3)]
+        public float Wind { get; set; }
+
+        /// <summary>
+        /// Precipitation
+        /// </summary>
+        [ColumnName("Rain"), LoadColumn(4)]
+        public float Rain { get; set; }
     }
 }

Server/ServerApp/Predictor/ModelOutput.cs

@@ -15 +15 @@
         /// <summary>
         /// A predicted class.
         /// </summary>
-        [ColumnName("prediction")]
+        [ColumnName("PredictedLabel")]
         public String Prediction { get; set; }
 
-        /// <summary>
-        /// The score of prediction probability into individual classes.
-        /// </summary>
-        public float[] Score { get; set; }
-
     }
 }

Server/ServerApp/Predictor/NaiveBayesClassifier.cs

@@ -35 +35 @@
         public void Fit(IEnumerable<ModelInput> trainInput)
         {
             IDataView trainingDataView = mlContext.Data.LoadFromEnumerable(trainInput);
-            var dataProcessPipeline = mlContext.Transforms.Conversion.MapValueToKey(nameof(ModelInput.Label))
-                .Append(mlContext.Transforms.Concatenate("Features", new[] { "temp" })
-                .Append(mlContext.Transforms.NormalizeMinMax("Features", "Features")));
-            var trainer = mlContext.MulticlassClassification.Trainers.NaiveBayes();
-            var traininingPipeline = dataProcessPipeline.Append(trainer)
-                .Append(mlContext.Transforms.Conversion.MapKeyToValue("prediction", "PredictedLabel"));
-
-            this.model = traininingPipeline.Fit(trainingDataView);
+            var pipeline = mlContext.Transforms.Conversion.MapValueToKey(nameof(ModelInput.Label))
+                .Append(mlContext.Transforms.Concatenate("Features", new[] { "Temp" }))
+                .Append(mlContext.Transforms.NormalizeMinMax("Features", "Features"))
+                .Append(mlContext.MulticlassClassification.Trainers.NaiveBayes());
+
+            this.model =pipeline.Fit(trainingDataView);
 
         }
 

Server/ServerApp/Predictor/PredictionController.cs

@@ -92 +92 @@
                 // TODO A single predictor is used for all areas, so training is done only once now.
                 for (int i = 0; i < this.predictors.Count; i++)
                 {
-                    // TODO change datetimes when parser interface is ready to parse only downloaded data.
-                    //IEnumerable<ModelInput> data = featureExtractor.PrepareModelInput(i, DateTime.MinValue, DateTime.MaxValue);
-                    IEnumerable<ModelInput> data = featureExtractor.PrepareModelInput(i, new DateTime(2019, 10, 5), new DateTime(2020, 6, 30));
+                    // train on all available data
+                    IEnumerable<ModelInput> data = featureExtractor.PrepareModelInput(i, DateTime.MinValue, DateTime.MaxValue);
                     this.predictors[i].Fit(data);
                 }
             } else

Server/ServerApp/Program.cs

@@ -84 +84 @@
 
 			// TODO nastavit čas
 			IDataParser p = new DataParser(dd);
-			DateTime startT = new DateTime(2019, 10, 5);
-			DateTime endT = new DateTime(2020, 6, 30);
             IPredictionController predictionController = new PredictionController(p);
             predictionController.Train();
             //var results = predictionController.Predict()