F# migrate to v1.0.0-preview (dotnet#354)

* F# - migrate bike sharing sample

* F# - migrate spam sample

* F# - migrate fraud sample

* F# - migrate customer seg sample

* F# - migrate iris classification sample

* F# - migrate iris cluster sample

* F# migrate sentiment sample

* F# migrate taxi sample

* rename solution

* F# migrate tensorflow image classification sample

* Label 'Count' column as 'Label'

* print full trainer name

* print absolute path

* sentiment data update, some formatting updates

* update tensorflow model

* F# migrate issue labeler

* update readme

* update readme

Author: kevmal

samples/fsharp/Directory.Build.props

@@ -1,7 +1,7 @@
 <Project>
 
   <PropertyGroup>
-    <MicrosoftMLVersion>0.11.0</MicrosoftMLVersion>
+    <MicrosoftMLVersion>1.0.0-preview</MicrosoftMLVersion>
   </PropertyGroup>
 
 </Project>

samples/fsharp/common/ConsoleHelper.fs

@@ -6,7 +6,6 @@ module ConsoleHelper =
     open Microsoft.ML.Data
     //open Microsoft.ML.Api
     open System.Reflection
-    open Microsoft.Data.DataView
 
     let printPrediction prediction =
         printfn "*************************************************"
@@ -23,20 +22,20 @@ module ConsoleHelper =
         printfn "*************************************************"
         printfn "*       Metrics for %s regression model      " name
         printfn "*------------------------------------------------"
-        printfn "*       LossFn:        %.2f" metrics.LossFn
+        printfn "*       LossFn:        %.2f" metrics.LossFunction
         printfn "*       R2 Score:      %.2f" metrics.RSquared
-        printfn "*       Absolute loss: %.2f" metrics.L1
-        printfn "*       Squared loss:  %.2f" metrics.L2
-        printfn "*       RMS loss:      %.2f" metrics.Rms
+        printfn "*       Absolute loss: %.2f" metrics.MeanAbsoluteError
+        printfn "*       Squared loss:  %.2f" metrics.MeanSquaredError
+        printfn "*       RMS loss:      %.2f" metrics.RootMeanSquaredError
         printfn "*************************************************"
 
     let printBinaryClassificationMetrics name (metrics : CalibratedBinaryClassificationMetrics) =
         printfn"************************************************************"
         printfn"*       Metrics for %s binary classification model      " name
         printfn"*-----------------------------------------------------------"
         printfn"*       Accuracy: %.2f%%" (metrics.Accuracy * 100.)
-        printfn"*       Auc:      %.2f%%" (metrics.Auc * 100.)
-        printfn"*       Auprc:    %.2f%%" (metrics.Auprc * 100.)
+        printfn"*       Area Under Curve:      %.2f%%" (metrics.AreaUnderRocCurve * 100.)
+        printfn"*       Area under Precision recall Curve:    %.2f%%" (metrics.AreaUnderPrecisionRecallCurve * 100.)
         printfn"*       F1Score:  %.2f%%" (metrics.F1Score * 100.)
 
         printfn"*       LogLogg:  %.2f%%" (metrics.LogLoss)
@@ -47,12 +46,12 @@ module ConsoleHelper =
         printfn"*       NegativeRecall:      %.2f" (metrics.NegativeRecall)
         printfn"************************************************************"
 
-    let printMultiClassClassificationMetrics name (metrics : MultiClassClassifierMetrics) =
+    let printMultiClassClassificationMetrics name (metrics : MulticlassClassificationMetrics) =
         printfn "************************************************************"
         printfn "*    Metrics for %s multi-class classification model   " name
         printfn "*-----------------------------------------------------------"
-        printfn "    AccuracyMacro = %.4f, a value between 0 and 1, the closer to 1, the better" metrics.AccuracyMacro
-        printfn "    AccuracyMicro = %.4f, a value between 0 and 1, the closer to 1, the better" metrics.AccuracyMicro
+        printfn "    AccuracyMacro = %.4f, a value between 0 and 1, the closer to 1, the better" metrics.MacroAccuracy
+        printfn "    AccuracyMicro = %.4f, a value between 0 and 1, the closer to 1, the better" metrics.MacroAccuracy
         printfn "    LogLoss = %.4f, the closer to 0, the better" metrics.LogLoss
         printfn "    LogLoss for class 1 = %.4f, the closer to 0, the better" metrics.PerClassLogLoss.[0]
         printfn "    LogLoss for class 2 = %.4f, the closer to 0, the better" metrics.PerClassLogLoss.[1]
@@ -70,16 +69,16 @@ module ConsoleHelper =
         let confidenceInterval95 = 1.96 * calculateStandardDeviation(values) / Math.Sqrt(float (values.Length-1));
         confidenceInterval95
 
-    let printMulticlassClassificationFoldsAverageMetrics algorithmName (crossValResults : (MultiClassClassifierMetrics * ITransformer * IDataView) array) =
+    let printMulticlassClassificationFoldsAverageMetrics algorithmName (crossValResults : TrainCatalogBase.CrossValidationResult<MulticlassClassificationMetrics>[]) =
 
-        let metricsInMultipleFolds = crossValResults |> Array.map(fun (metrics, model, scoredTestData) -> metrics)
+        let metricsInMultipleFolds = crossValResults |> Array.map(fun r -> r.Metrics)
 
-        let microAccuracyValues  = metricsInMultipleFolds |> Array.map(fun m -> m.AccuracyMicro)
+        let microAccuracyValues  = metricsInMultipleFolds |> Array.map(fun m -> m.MicroAccuracy)
         let microAccuracyAverage = microAccuracyValues |> Array.average
         let microAccuraciesStdDeviation = calculateStandardDeviation microAccuracyValues
         let microAccuraciesConfidenceInterval95 = calculateConfidenceInterval95 microAccuracyValues
 
-        let macroAccuracyValues = metricsInMultipleFolds |> Array.map(fun m -> m.AccuracyMacro)
+        let macroAccuracyValues = metricsInMultipleFolds |> Array.map(fun m -> m.MicroAccuracy)
         let macroAccuracyAverage = macroAccuracyValues |> Array.average
         let macroAccuraciesStdDeviation = calculateStandardDeviation macroAccuracyValues
         let macroAccuraciesConfidenceInterval95 = calculateConfidenceInterval95 macroAccuracyValues
@@ -107,8 +106,8 @@ module ConsoleHelper =
         printfn "*************************************************"
         printfn "*       Metrics for %s clustering model      " name
         printfn "*------------------------------------------------"
-        printfn "*       AvgMinScore: %.15f" metrics.AvgMinScore
-        printfn "*       DBI is: %.15f" metrics.Dbi
+        printfn "*       Average Distance: %.15f" metrics.AverageDistance
+        printfn "*       Davies Bouldin Index is: %.15f" metrics.DaviesBouldinIndex
         printfn "*************************************************"
 
     let consoleWriteHeader line =
@@ -157,7 +156,7 @@ module ConsoleHelper =
 
         // Extract the 'Features' column.
         let someColumnData = 
-            transformedData.GetColumn<float32[]>(mlContext, columnName)
+            transformedData.GetColumn<float32[]>(columnName)
             |> Seq.take numberOfRows
             |> Seq.toList
 

samples/fsharp/end-to-end-apps/MulticlassClassification-GitHubLabeler/GitHubLabeler/GitHubLabelerConsoleApp/GitHubLabeler.fsproj

@@ -23,7 +23,7 @@
   <ItemGroup>
     <PackageReference Include="Microsoft.Extensions.Configuration" Version="2.1.1" />
     <PackageReference Include="Microsoft.Extensions.Configuration.Json" Version="2.1.1" />
-    <PackageReference Include="Microsoft.ML" Version="0.11.0" />
+    <PackageReference Include="Microsoft.ML" Version="$(MicrosoftMLVersion)" />
     <PackageReference Include="Octokit" Version="0.29.0" />
   </ItemGroup>
 

samples/fsharp/end-to-end-apps/MulticlassClassification-GitHubLabeler/GitHubLabeler/GitHubLabelerConsoleApp/Labeler.fs

@@ -37,14 +37,14 @@ type GitHubClientFacade =
 let initialise modelPath repoOwner repoName accessToken =
     let mlContext = MLContext(seed = Nullable 1)
 
-    let trainedModel = 
+    let trainedModel, inputSchema = 
         use f = IO.File.OpenRead(modelPath)
         mlContext.Model.Load(f)
 
     let productInformation = ProductHeaderValue "MLGitHubLabeler"
     let client = GitHubClient(productInformation, Credentials = Credentials(accessToken))
     let gitHubClient = GitHubClientFacade.init client repoOwner repoName
-    let predictionEngine = trainedModel.CreatePredictionEngine<GitHubIssue, GitHubIssuePrediction>(mlContext)
+    let predictionEngine = mlContext.Model.CreatePredictionEngine<GitHubIssue, GitHubIssuePrediction>(trainedModel)
     predictionEngine, gitHubClient
 
 type FullPrediction = 

samples/fsharp/end-to-end-apps/MulticlassClassification-GitHubLabeler/GitHubLabeler/GitHubLabelerConsoleApp/Program.fs

@@ -76,17 +76,11 @@ let buildAndTrainModel dataSetLocation modelPath selectedStrategy =
     let trainer =
         match selectedStrategy with
         | MyTrainerStrategy.SdcaMultiClassTrainer ->
-            mlContext.MulticlassClassification.Trainers.StochasticDualCoordinateAscent(
-                DefaultColumnNames.Label, 
-                DefaultColumnNames.Features)
-            |> downcastPipeline
+            mlContext.MulticlassClassification.Trainers.SdcaMaximumEntropy( "Label", "Features") |> downcastPipeline
 
         | MyTrainerStrategy.OVAAveragedPerceptronTrainer ->
             let averagedPerceptronBinaryTrainer = 
-                mlContext.BinaryClassification.Trainers.AveragedPerceptron(
-                    DefaultColumnNames.Label,
-                    DefaultColumnNames.Features,
-                    numIterations = 10)
+                mlContext.BinaryClassification.Trainers.AveragedPerceptron( "Label", "Features", numberOfIterations = 10)
 
             let downcastTrainer (x : ITrainerEstimator<_,_>) = 
                 match x with 
@@ -117,16 +111,16 @@ let buildAndTrainModel dataSetLocation modelPath selectedStrategy =
     let watchCrossValTime = System.Diagnostics.Stopwatch.StartNew()
 
     let crossValidationResults = 
-        mlContext.MulticlassClassification.CrossValidate(data = trainingDataView, estimator = downcastPipeline modelBuilder, numFolds = 6, labelColumn = DefaultColumnNames.Label)
+        mlContext.MulticlassClassification.CrossValidate(data = trainingDataView, estimator = downcastPipeline modelBuilder, numberOfFolds = 6, labelColumnName = "Label")
 
 
     //Stop measuring time
     watchCrossValTime.Stop()
     printfn "Time Cross-Validating: %d miliSecs"  watchCrossValTime.ElapsedMilliseconds
 
     crossValidationResults
-    |> Array.map (fun x -> x.Metrics, x.Model, x.ScoredHoldOutSet) //convert struct tuple for print function
-    |> Common.ConsoleHelper.printMulticlassClassificationFoldsAverageMetrics (trainer.ToString())
+    |> Seq.toArray
+    |> Common.ConsoleHelper.printMulticlassClassificationFoldsAverageMetrics (trainer.ToString()) 
 
     // STEP 5: Train the model fitting to the DataSet
     printfn "=============== Training the model ==============="
@@ -140,7 +134,7 @@ let buildAndTrainModel dataSetLocation modelPath selectedStrategy =
         Title = "WebSockets communication is slow in my machine"
         Description = "The WebSockets communication used under the covers by SignalR looks like is going slow in my development machine.." 
     }
-    let predEngine = trainedModel.CreatePredictionEngine<GitHubIssue, GitHubIssuePrediction>(mlContext)
+    let predEngine = mlContext.Model.CreatePredictionEngine<GitHubIssue, GitHubIssuePrediction>(trainedModel)
     let prediction =  predEngine.Predict(issue)
 
     printfn "=============== Single Prediction just-trained-model - Result: %s ===============" prediction.Area
@@ -149,7 +143,7 @@ let buildAndTrainModel dataSetLocation modelPath selectedStrategy =
     printfn "=============== Saving the model to a file ==============="
     do 
         use f = File.Open(modelPath,FileMode.Create)
-        mlContext.Model.Save(trainedModel, f)
+        mlContext.Model.Save(trainedModel, trainingDataView.Schema, f)
 
     Common.ConsoleHelper.consoleWriteHeader "Training process finalized"
 

samples/fsharp/end-to-end-apps/MulticlassClassification-GitHubLabeler/README.md

@@ -2,7 +2,7 @@
 
 | ML.NET version | API type          | Status                        | App Type    | Data sources | Scenario            | ML Task                   | Algorithms                  |
 |----------------|-------------------|-------------------------------|-------------|-----------|---------------------|---------------------------|-----------------------------|
-| v0.11           | Dynamic API | Up-to-date | Console app | .csv file and GitHub issues | Issues classification | Multi-class  classification | SDCA multi-class classifier |
+| v1.0.0-preview | Dynamic API | Up-to-date | Console app | .csv file and GitHub issues | Issues classification | Multi-class  classification | SDCA multi-class classifier |
 
 
 This is a simple prototype application to demonstrate how to use [ML.NET](https://www.nuget.org/packages/Microsoft.ML/) APIs. The main focus is on creating, training, and using ML (Machine Learning) model that is implemented in Predictor.cs class.

samples/fsharp/getting-started/BinaryClassification_CreditCardFraudDetection/CreditCardFraudDetection/CreditCardFraudDetection.fsproj

@@ -11,6 +11,7 @@
 
   <ItemGroup>
     <PackageReference Include="Microsoft.ML" Version="$(MicrosoftMLVersion)" />
+    <PackageReference Include="Microsoft.ML.FastTree" Version="1.0.0-preview" />
   </ItemGroup>
 
 </Project>

samples/fsharp/getting-started/BinaryClassification_CreditCardFraudDetection/CreditCardFraudDetection/Program.fs

@@ -96,7 +96,7 @@ let main _ =
         let data = mlContext.Data.LoadFromTextFile<TransactionObservation>(inputFile, separatorChar = ',', hasHeader = true, allowQuoting = true)
 
         let trainData, testData = 
-            let y = mlContext.BinaryClassification.TrainTestSplit(data, 0.2, seed = Nullable 1u) 
+            let y = mlContext.Data.TrainTestSplit(data, 0.2, seed = Nullable 1) 
             y.TrainSet, y.TestSet
 
         // save test split
@@ -132,24 +132,23 @@ let main _ =
 
     let pipeline = 
         EstimatorChain()
-        |> fun x -> x.Append(mlContext.Transforms.Concatenate(DefaultColumnNames.Features, featureColumnNames))
+        |> fun x -> x.Append(mlContext.Transforms.Concatenate("Features", featureColumnNames))
         |> fun x -> x.Append(mlContext.Transforms.DropColumns [|"Time"|])
         |> fun x -> 
             x.Append (
-                mlContext.Transforms.Normalize (
+                mlContext.Transforms.NormalizeMeanVariance (
                     "FeaturesNormalizedByMeanVar", 
-                    "Features", 
-                    NormalizingEstimator.NormalizerMode.MeanVariance
+                    "Features"
                     )
                 )
         |> fun x -> 
             x.Append (
                 mlContext.BinaryClassification.Trainers.FastTree(
                     "Label", 
                     "FeaturesNormalizedByMeanVar", 
-                    numLeaves = 20, 
-                    numTrees = 100, 
-                    minDatapointsInLeaves = 10, 
+                    numberOfLeaves = 20, 
+                    numberOfTrees = 100, 
+                    minimumExampleCountPerLeaf = 10, 
                     learningRate = 0.2
                     )
                 )
@@ -163,18 +162,18 @@ let main _ =
     printfn "Saving model to file"
     let _ = 
         use fs = new FileStream (modelFile, FileMode.Create, FileAccess.Write, FileShare.Write)
-        mlContext.Model.Save(model, fs)
+        mlContext.Model.Save(model, trainData.Schema, fs)
 
     (*
     Read the model and test data from file,
     and make predictions
     *)
 
     printfn "Reading model and test data"
-    let modelEvaluator = 
+    let modelEvaluator, inputSchema = 
         use file = File.OpenRead modelFile           
         mlContext.Model.Load(file)
-    let predictionEngine = modelEvaluator.CreatePredictionEngine<TransactionObservation, TransactionFraudPrediction>(mlContext)
+    let predictionEngine = mlContext.Model.CreatePredictionEngine<TransactionObservation, TransactionFraudPrediction>(modelEvaluator)
 
     let testData = mlContext.Data.LoadFromTextFile<TransactionObservation>(testFile, hasHeader = true, separatorChar = ',')
 

samples/fsharp/getting-started/BinaryClassification_CreditCardFraudDetection/Readme.md

@@ -2,7 +2,7 @@
 
 | ML.NET version | API type          | Status                        | App Type    | Data type | Scenario            | ML Task                   | Algorithms                  |
 |----------------|-------------------|-------------------------------|-------------|-----------|---------------------|---------------------------|-----------------------------|
-| v0.11           | Dynamic API | Up-to-date | Two console apps | .csv file | Fraud Detection | Two-class classification | FastTree Binary Classification |
+| v1.0.0-preview | Dynamic API | Up-to-date | Two console apps | .csv file | Fraud Detection | Two-class classification | FastTree Binary Classification |
 
 In this introductory sample, you'll see how to use ML.NET to predict a credit card fraud. In the world of machine learning, this type of prediction is known as binary classification.
 
@@ -69,38 +69,40 @@ The initial code is similar to the following:
 [...]
 
     let trainData, testData = 
-        let y = mlContext.BinaryClassification.TrainTestSplit(data, 0.2, seed = Nullable 1u) 
-        y.TrainSet, y.TestSet
+        printfn "Reading train and test data"
+        let trainData = mlContext.Data.LoadFromTextFile<TransactionObservation>(trainFile, separatorChar = ',', hasHeader = true)
+        let testData = mlContext.Data.LoadFromTextFile<TransactionObservation>(testFile, separatorChar = ',', hasHeader = true)
+        trainData, testData
 
 [...]
 
     let featureColumnNames = 
         trainData.Schema
         |> Seq.map (fun column -> column.Name)
+        |> Seq.filter (fun name -> name <> "Time")
         |> Seq.filter (fun name -> name <> "Label")
-        |> Seq.filter (fun name -> name <> "StratificationColumn")
+        |> Seq.filter (fun name -> name <> "IdPreservationColumn")
         |> Seq.toArray
 
     let pipeline = 
         EstimatorChain()
-        |> fun x -> x.Append(mlContext.Transforms.Concatenate(DefaultColumnNames.Features, featureColumnNames))
+        |> fun x -> x.Append(mlContext.Transforms.Concatenate("Features", featureColumnNames))
         |> fun x -> x.Append(mlContext.Transforms.DropColumns [|"Time"|])
         |> fun x -> 
             x.Append (
-                mlContext.Transforms.Normalize (
+                mlContext.Transforms.NormalizeMeanVariance (
                     "FeaturesNormalizedByMeanVar", 
-                    "Features", 
-                    NormalizingEstimator.NormalizerMode.MeanVariance
+                    "Features"
                     )
                 )
         |> fun x -> 
             x.Append (
                 mlContext.BinaryClassification.Trainers.FastTree(
                     "Label", 
                     "FeaturesNormalizedByMeanVar", 
-                    numLeaves = 20, 
-                    numTrees = 100, 
-                    minDatapointsInLeaves = 10, 
+                    numberOfLeaves = 20, 
+                    numberOfTrees = 100, 
+                    minimumExampleCountPerLeaf = 10, 
                     learningRate = 0.2
                     )
                 )
@@ -130,12 +132,13 @@ After the model is trained, you can use the `Predict()` API to predict if a tran
 
 `````fsharp
     printfn "Making predictions"
-    mlContext.CreateEnumerable<TransactionObservation>(testData, reuseRowObject = false)
+    mlContext.Data.CreateEnumerable<TransactionObservation>(testData, reuseRowObject = false)
     |> Seq.filter (fun x -> x.Label = true)
     // use 5 observations from the test data
     |> Seq.take 5
     |> Seq.iter (fun testData -> 
         let prediction = predictionEngine.Predict testData
         printfn "%A" prediction
         printfn "------"
+        )
 `````