Use transformer scopes to ignore labels when scoring

TransformerScope.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Install packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div><div></div><div></div><div><strong>Installed Packages</strong><ul><li><span>Microsoft.ML, 2.0.0-preview.22551.1</span></li></ul></div></div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#r \"nuget: Microsoft.ML, 2.0.0-preview.22551.1\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Add using statements"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "using Microsoft.ML;\n",
    "using Microsoft.ML.Data;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Initialize MLContext"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var ctx = new MLContext();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define text loader options"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var options = new TextLoader.Options\n",
    "{\n",
    "    Separators=new [] {','},\n",
    "    HasHeader=false,\n",
    "    Columns=new [] {\n",
    "        new TextLoader.Column(\"Features\",DataKind.Single,0,3),\n",
    "        new TextLoader.Column(\"Label\",DataKind.String,4)\n",
    "    }\n",
    "};"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create text loader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var textLoader = ctx.Data.CreateTextLoader(options);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load data into IDataView"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var data = textLoader.Load(@\"C:\\Datasets\\iris.data.txt\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preview schema"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table><thead><tr><th><i>index</i></th><th>Name</th><th>Index</th><th>IsHidden</th><th>Type</th><th>Annotations</th></tr></thead><tbody><tr><td>0</td><td>Features</td><td><div class=\"dni-plaintext\">0</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>Dimensions</th><th>IsKnownSize</th><th>ItemType</th><th>Size</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 4 ]</div></td><td><div class=\"dni-plaintext\">True</div></td><td><div class=\"dni-plaintext\">{ Single: RawType: System.Single }</div></td><td><div class=\"dni-plaintext\">4</div></td><td><div class=\"dni-plaintext\">Microsoft.ML.Data.VBuffer&lt;System.Single&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[  ]</div></td></tr></tbody></table></td></tr><tr><td>1</td><td>Label</td><td><div class=\"dni-plaintext\">1</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">System.ReadOnlyMemory&lt;System.Char&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[  ]</div></td></tr></tbody></table></td></tr></tbody></table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data.Schema"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define training pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var pipeline = \n",
    "    ctx.Transforms.NormalizeMinMax(\"NormalizedFeatures\",\"Features\")\n",
    "        .Append(ctx.Transforms.Conversion.MapValueToKey(\"Label\"), TransformerScope.Training)\n",
    "        .Append(ctx.MulticlassClassification.Trainers.SdcaMaximumEntropy())\n",
    "        .Append(ctx.Transforms.Conversion.MapKeyToValue(\"PredictedLabel\"));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var model = pipeline.Fit(data);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using Transform (automatic)\n",
    "\n",
    "Calling `Transform` automatically sets the scope to scoring."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Remove Label column from IDataView"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table><thead><tr><th><i>index</i></th><th>Name</th><th>Index</th><th>IsHidden</th><th>Type</th><th>Annotations</th></tr></thead><tbody><tr><td>0</td><td>Features</td><td><div class=\"dni-plaintext\">0</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>Dimensions</th><th>IsKnownSize</th><th>ItemType</th><th>Size</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 4 ]</div></td><td><div class=\"dni-plaintext\">True</div></td><td><div class=\"dni-plaintext\">{ Single: RawType: System.Single }</div></td><td><div class=\"dni-plaintext\">4</div></td><td><div class=\"dni-plaintext\">Microsoft.ML.Data.VBuffer&lt;System.Single&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[  ]</div></td></tr></tbody></table></td></tr></tbody></table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "var idvWithoutLabels = \n",
    "    ctx.Transforms.DropColumns(\"Label\")\n",
    "        .Fit(data)\n",
    "        .Transform(data);\n",
    "\n",
    "idvWithoutLabels.Schema"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make batch predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var batchPredictions = model.Transform(idvWithoutLabels);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preview batch prediction schema"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table><thead><tr><th><i>index</i></th><th>Name</th><th>Index</th><th>IsHidden</th><th>Type</th><th>Annotations</th></tr></thead><tbody><tr><td>0</td><td>Features</td><td><div class=\"dni-plaintext\">0</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>Dimensions</th><th>IsKnownSize</th><th>ItemType</th><th>Size</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 4 ]</div></td><td><div class=\"dni-plaintext\">True</div></td><td><div class=\"dni-plaintext\">{ Single: RawType: System.Single }</div></td><td><div class=\"dni-plaintext\">4</div></td><td><div class=\"dni-plaintext\">Microsoft.ML.Data.VBuffer&lt;System.Single&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[  ]</div></td></tr></tbody></table></td></tr><tr><td>1</td><td>NormalizedFeatures</td><td><div class=\"dni-plaintext\">1</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>Dimensions</th><th>IsKnownSize</th><th>ItemType</th><th>Size</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 4 ]</div></td><td><div class=\"dni-plaintext\">True</div></td><td><div class=\"dni-plaintext\">{ Single: RawType: System.Single }</div></td><td><div class=\"dni-plaintext\">4</div></td><td><div class=\"dni-plaintext\">Microsoft.ML.Data.VBuffer&lt;System.Single&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ { IsNormalized: Boolean: Name: IsNormalized, Index: 0, IsHidden: False, Type: { Boolean: RawType: System.Boolean }, Annotations: { : Schema: [  ] } } ]</div></td></tr></tbody></table></td></tr><tr><td>2</td><td>PredictedLabel</td><td><div class=\"dni-plaintext\">2</div></td><td><div class=\"dni-plaintext\">True</div></td><td><table><thead><tr><th>Count</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">3</div></td><td><div class=\"dni-plaintext\">System.UInt32</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ { ScoreColumnKind: String: Name: ScoreColumnKind, Index: 0, IsHidden: False, Type: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Annotations: { : Schema: [  ] } }, { ScoreValueKind: String: Name: ScoreValueKind, Index: 1, IsHidden: False, Type: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Annotations: { : Schema: [  ] } }, { KeyValues: Vector&lt;String, 3&gt;: Name: KeyValues, Index: 2, IsHidden: False, Type: { Vector&lt;String, 3&gt;: Dimensions: [ 3 ], IsKnownSize: True, ItemType: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Size: 3, RawType: Microsoft.ML.Data.VBuffer&lt;System.ReadOnlyMemory&lt;System.Char&gt;&gt; }, Annotations: { : Schema: [  ] } }, { ScoreColumnSetId: Key&lt;UInt32, 0-2147483646&gt;: Name: ScoreColumnSetId, Index: 3, IsHidden: False, Type: { Key&lt;UInt32, 0-2147483646&gt;: Count: 2147483647, RawType: System.UInt32 }, Annotations: { : Schema: [  ] } } ]</div></td></tr></tbody></table></td></tr><tr><td>3</td><td>PredictedLabel</td><td><div class=\"dni-plaintext\">3</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">System.ReadOnlyMemory&lt;System.Char&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[  ]</div></td></tr></tbody></table></td></tr><tr><td>4</td><td>Score</td><td><div class=\"dni-plaintext\">4</div></td><td><div class=\"dni-plaintext\">False</div></td><td><table><thead><tr><th>Dimensions</th><th>IsKnownSize</th><th>ItemType</th><th>Size</th><th>RawType</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 3 ]</div></td><td><div class=\"dni-plaintext\">True</div></td><td><div class=\"dni-plaintext\">{ Single: RawType: System.Single }</div></td><td><div class=\"dni-plaintext\">3</div></td><td><div class=\"dni-plaintext\">Microsoft.ML.Data.VBuffer&lt;System.Single&gt;</div></td></tr></tbody></table></td><td><table><thead><tr><th>Schema</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ { ScoreColumnKind: String: Name: ScoreColumnKind, Index: 0, IsHidden: False, Type: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Annotations: { : Schema: [  ] } }, { ScoreColumnSetId: Key&lt;UInt32, 0-2147483646&gt;: Name: ScoreColumnSetId, Index: 1, IsHidden: False, Type: { Key&lt;UInt32, 0-2147483646&gt;: Count: 2147483647, RawType: System.UInt32 }, Annotations: { : Schema: [  ] } }, { ScoreValueKind: String: Name: ScoreValueKind, Index: 2, IsHidden: False, Type: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Annotations: { : Schema: [  ] } }, { TrainingLabelValues: Vector&lt;String, 3&gt;: Name: TrainingLabelValues, Index: 3, IsHidden: False, Type: { Vector&lt;String, 3&gt;: Dimensions: [ 3 ], IsKnownSize: True, ItemType: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Size: 3, RawType: Microsoft.ML.Data.VBuffer&lt;System.ReadOnlyMemory&lt;System.Char&gt;&gt; }, Annotations: { : Schema: [  ] } }, { SlotNames: Vector&lt;String, 3&gt;: Name: SlotNames, Index: 4, IsHidden: False, Type: { Vector&lt;String, 3&gt;: Dimensions: [ 3 ], IsKnownSize: True, ItemType: { String: RawType: System.ReadOnlyMemory&lt;System.Char&gt; }, Size: 3, RawType: Microsoft.ML.Data.VBuffer&lt;System.ReadOnlyMemory&lt;System.Char&gt;&gt; }, Annotations: { : Schema: [  ] } } ]</div></td></tr></tbody></table></td></tr></tbody></table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "batchPredictions.Schema"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using PredictionEngine"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set scope to scoring\n",
    "\n",
    "This allows you to not pass in the label column."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var scoringModel = model.GetModelFor(TransformerScope.Scoring);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define model input and output schema"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "public class ModelInput\n",
    "{\n",
    "    [VectorType(4)]\n",
    "    public float[] Features {get;set;}\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "public class ModelOutput\n",
    "{\n",
    "    [VectorType(3)]\n",
    "    public float[] Score {get;set;}\n",
    "    \n",
    "    public string PredictedLabel{get;set;} \n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create prediction engine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var predictionEngine = \n",
    "    ctx.Model.CreatePredictionEngine<ModelInput,ModelOutput>(scoringModel);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define sample input to predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var input = new ModelInput {Features = new [] {5.1f,3.5f,1.4f,0.2f}};"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make single prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [],
   "source": [
    "var prediction = predictionEngine.Predict(input);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### View prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "dotnet_interactive": {
     "language": "csharp"
    },
    "vscode": {
     "languageId": "dotnet-interactive.csharp"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table><thead><tr><th>Score</th><th>PredictedLabel</th></tr></thead><tbody><tr><td><div class=\"dni-plaintext\">[ 0.9999666, 3.3485834E-05, 2.356079E-22 ]</div></td><td><div class=\"dni-plaintext\">Iris-setosa</div></td></tr></tbody></table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "prediction"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".NET (C#)",
   "language": "C#",
   "name": ".net-csharp"
  },
  "language_info": {
   "name": "C#"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}