Multiclass nondefault label test (dotnet#208)

Dmitry-A · Feb 23, 2019 · abf6f0d · abf6f0d
1 parent 796ae8a
commit abf6f0d
Show file tree

Hide file tree

Showing 4 changed files with 190 additions and 8 deletions.
diff --git a/src/Test/AutoFitTests.cs b/src/Test/AutoFitTests.cs
@@ -31,15 +31,14 @@ public void AutoFitBinaryTest()
         public void AutoFitMultiTest()
         {
             var context = new MLContext();
-            var dataPath = DatasetUtil.DownloadTrivialDataset();
-            var columnInference = context.AutoInference().InferColumns(dataPath, DatasetUtil.TrivialDatasetLabel);
+            var columnInference = context.AutoInference().InferColumns(DatasetUtil.TrivialMulticlassDatasetPath, DatasetUtil.TrivialMulticlassDatasetLabel);
             var textLoader = context.Data.CreateTextLoader(columnInference.TextLoaderArgs);
-            var trainData = textLoader.Read(dataPath);
+            var trainData = textLoader.Read(DatasetUtil.TrivialMulticlassDatasetPath);
             var validationData = context.Data.TakeRows(trainData, 20);
             trainData = context.Data.SkipRows(trainData, 20);
             var result = context.AutoInference()
                 .CreateMulticlassClassificationExperiment(0)
-                .Execute(trainData, validationData, new ColumnInformation() { LabelColumn = DatasetUtil.TrivialDatasetLabel });
+                .Execute(trainData, validationData, new ColumnInformation() { LabelColumn = DatasetUtil.TrivialMulticlassDatasetLabel });
 
             Assert.IsTrue(result.Max(i => i.Metrics.AccuracyMacro) > 0.80);
         }

diff --git a/src/Test/DatasetUtil.cs b/src/Test/DatasetUtil.cs
@@ -13,10 +13,12 @@ namespace Microsoft.ML.Auto.Test
     internal static class DatasetUtil
     {
         public const string UciAdultLabel = DefaultColumnNames.Label;
-        public const string TrivialDatasetLabel = DefaultColumnNames.Label;
+        public const string TrivialMulticlassDatasetLabel = "Target";
         public const string MlNetGeneratedRegressionLabel = "target";
         public const int IrisDatasetLabelColIndex = 0;
 
+        public const string TrivialMulticlassDatasetPath = @"TestData\TrivialMulticlassDataset.txt";
+
         private static IDataView _uciAdultDataView;
 
         public static IDataView GetUciAdultDataView()
@@ -36,9 +38,6 @@ public static IDataView GetUciAdultDataView()
         public static string DownloadUciAdultDataset() =>
             DownloadIfNotExists("https://raw.githubusercontent.com/dotnet/machinelearning/f0e639af5ffdc839aae8e65d19b5a9a1f0db634a/test/data/adult.tiny.with-schema.txt", "uciadult.dataset");
 
-        public static string DownloadTrivialDataset() =>
-            DownloadIfNotExists("https://raw.githubusercontent.com/dotnet/machinelearning/eae76959e6714af44caa212e102a5f06f0110e72/test/data/trivial-train.tsv", "trivial.dataset");
-
         public static string DownloadMlNetGeneratedRegressionDataset() =>
             DownloadIfNotExists("https://raw.githubusercontent.com/dotnet/machinelearning/e78971ea6fd736038b4c355b840e5cbabae8cb55/test/data/generated_regression_dataset.csv", "mlnet_generated_regression.dataset");
 

diff --git a/src/Test/Test.csproj b/src/Test/Test.csproj
@@ -31,6 +31,9 @@
     <None Update="TestData\DatasetWithEmptyColumn.txt">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
+    <None Update="TestData\TrivialMulticlassDataset.txt">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
   </ItemGroup>
 
 </Project>
diff --git a/src/Test/TestData/TrivialMulticlassDataset.txt b/src/Test/TestData/TrivialMulticlassDataset.txt
@@ -0,0 +1,181 @@
+Target	Row	Column
+1	14	20
+1	19	26
+3	17	4
+1	10	20
+1	3	5
+1	7	5
+1	18	36
+2	1	36
+2	1	38
+3	17	1
+2	6	26
+2	9	30
+3	13	8
+2	7	33
+2	8	30
+3	10	1
+1	18	25
+1	13	12
+1	3	2
+2	8	28
+1	11	24
+2	3	28
+2	1	16
+1	9	7
+1	15	16
+3	19	4
+1	1	8
+1	8	0
+1	10	34
+1	18	37
+2	1	17
+2	8	39
+1	17	30
+2	1	27
+2	0	38
+1	11	16
+3	19	3
+1	7	8
+1	13	13
+1	19	31
+3	16	1
+1	5	1
+2	6	11
+1	9	5
+3	10	6
+1	1	2
+2	6	30
+2	7	15
+1	17	21
+1	18	23
+3	10	7
+2	5	39
+2	2	27
+3	12	6
+3	11	4
+1	9	3
+1	12	22
+2	8	19
+2	1	14
+1	11	11
+1	10	36
+3	12	4
+1	15	21
+1	17	37
+1	6	3
+2	3	18
+1	10	10
+1	11	33
+1	18	19
+2	7	35
+3	10	2
+1	12	30
+1	12	26
+2	1	31
+2	5	21
+2	1	11
+1	7	3
+2	8	36
+3	10	4
+1	18	26
+2	8	10
+1	10	22
+1	15	14
+3	16	0
+2	0	30
+2	3	34
+3	13	9
+1	0	2
+1	15	36
+1	15	23
+1	10	30
+2	6	20
+2	9	24
+2	9	35
+1	7	6
+2	7	39
+2	5	20
+3	12	8
+2	9	12
+1	17	25
+1	12	33
+2	6	19
+1	17	10
+2	4	35
+1	15	31
+3	12	7
+1	17	16
+2	1	19
+2	3	25
+1	16	30
+1	19	30
+1	5	4
+2	6	10
+1	18	20
+1	13	26
+2	3	39
+2	2	20
+1	4	7
+2	3	33
+1	16	20
+2	1	21
+3	15	2
+3	19	2
+1	12	10
+2	5	37
+2	1	32
+3	18	6
+1	2	1
+1	16	21
+2	1	23
+1	17	33
+2	5	11
+2	3	14
+1	11	12
+1	13	20
+1	19	38
+1	15	10
+2	8	11
+3	11	0
+1	18	10
+1	19	24
+1	13	11
+2	4	23
+1	16	26
+1	7	7
+1	17	29
+1	18	30
+1	13	10
+2	6	21
+1	19	32
+2	7	12
+1	12	28
+2	2	11
+1	12	15
+2	8	32
+3	15	9
+3	16	5
+1	9	1
+1	19	28
+3	16	3
+1	15	17
+2	7	38
+1	16	38
+1	14	26
+1	10	26
+1	10	37
+3	18	5
+2	5	27
+2	2	22
+1	11	39
+1	16	36
+1	0	9
+2	5	19
+1	18	28
+1	12	13
+1	17	17
+1	8	1
+2	6	15
+3	14	4
+1	1	4