Remove unnecessary training data from Fit() methods (#1129)

Author: natke

machine-learning/tutorials/ProductSalesAnomalyDetection/Data/product-sales.csv

@@ -0,0 +1,37 @@
+Month,ProductSales
+1-Jan,271
+2-Jan,150.9
+3-Jan,188.1
+4-Jan,124.3
+5-Jan,185.3
+6-Jan,173.5
+7-Jan,236.8
+8-Jan,229.5
+9-Jan,197.8
+10-Jan,127.9
+11-Jan,341.5
+12-Jan,190.9
+1-Feb,199.3
+2-Feb,154.5
+3-Feb,215.1
+4-Feb,278.3
+5-Feb,196.4
+6-Feb,292
+7-Feb,231
+8-Feb,308.6
+9-Feb,294.9
+10-Feb,426.6
+11-Feb,269.5
+12-Feb,347.3
+1-Mar,344.7
+2-Mar,445.4
+3-Mar,320.9
+4-Mar,444.3
+5-Mar,406.3
+6-Mar,442.4
+7-Mar,580.5
+8-Mar,412.6
+9-Mar,687
+10-Mar,480.3
+11-Mar,586.3
+12-Mar,651.9

machine-learning/tutorials/ProductSalesAnomalyDetection/Program.cs

@@ -2,6 +2,7 @@
 using System;
 using System.IO;
 using Microsoft.ML;
+using System.Collections.Generic;
 // </SnippetAddUsings>
 
 namespace ProductSalesAnomalyDetection
@@ -44,17 +45,17 @@ static void DetectSpike(MLContext mlContext, int docSize, IDataView productSales
             var iidSpikeEstimator = mlContext.Transforms.DetectIidSpike(outputColumnName: nameof(ProductSalesPrediction.Prediction), inputColumnName: nameof(ProductSalesData.numSales), confidence: 95, pvalueHistoryLength: docSize / 4);
             // </SnippetAddSpikeTrainer> 
 
-            // STEP 3:Train the model by fitting the dataview
-            // Create and train the model based on the dataset that has been loaded, transformed.
+            // STEP 3: Create the transform
+            // Create the spike detection transform
             Console.WriteLine("=============== Training the model ===============");
             // <SnippetTrainModel1>
-            ITransformer trainedModel = iidSpikeEstimator.Fit(productSales);
+            ITransformer iidSpikeTransform = iidSpikeEstimator.Fit(CreateEmptyDataView(mlContext));
             // </SnippetTrainModel1>
 
             Console.WriteLine("=============== End of training process ===============");
              //Apply data transformation to create predictions.
             // <SnippetTransformData1>
-            IDataView transformedData = trainedModel.Transform(productSales);
+            IDataView transformedData = iidSpikeTransform.Transform(productSales);
             // </SnippetTransformData1>
 
             // <SnippetCreateEnumerable1>
@@ -90,16 +91,16 @@ static void DetectChangepoint(MLContext mlContext, int docSize, IDataView produc
             var iidChangePointEstimator = mlContext.Transforms.DetectIidChangePoint(outputColumnName: nameof(ProductSalesPrediction.Prediction), inputColumnName: nameof(ProductSalesData.numSales), confidence: 95, changeHistoryLength: docSize / 4);
             // </SnippetAddChangePointTrainer> 
 
-            //STEP 3:Train the model by fitting the dataview
+            //STEP 3: Create the transform
             Console.WriteLine("=============== Training the model Using Change Point Detection Algorithm===============");
             // <SnippetTrainModel2>
-            var trainedModel = iidChangePointEstimator.Fit(productSales);
+            var iidChangePointTransform = iidChangePointEstimator.Fit(CreateEmptyDataView(mlContext));
             // </SnippetTrainModel2>
             Console.WriteLine("=============== End of training process ===============");
 
             //Apply data transformation to create predictions.
             // <SnippetTransformData2>
-            IDataView transformedData = trainedModel.Transform(productSales);
+            IDataView transformedData = iidChangePointTransform.Transform(productSales);
             // </SnippetTransformData2>
 
             // <SnippetCreateEnumerable2>
@@ -124,5 +125,13 @@ static void DetectChangepoint(MLContext mlContext, int docSize, IDataView produc
             Console.WriteLine("");
             // </SnippetDisplayResults2>
         }
+
+        // <SnippetCreateEmptyDataView>
+        static IDataView CreateEmptyDataView(MLContext mlContext) {
+            // Create empty DataView. We just need the schema to call Fit() for the time series transforms
+            IEnumerable<ProductSalesData> enumerableData = new List<ProductSalesData>();
+            return mlContext.Data.LoadFromEnumerable(enumerableData);
+        }
+        // </SnippetCreateEmptyDataView>
     }
 }