ML model improvement : Added SHOW MODELS and DESCRIBE MODEL (#185)

* ML model improvement : Adding "SHOW MODELS and DESCRIBE MODEL"

Author:    rajagurunath <[email protected]>
Date:      Mon May 24 02:37:40 2021 +0530

* fix typo

* ML model improvement : refactoring for PR

* ML model improvement : Adding stmts in notebook

* ML model improvement : Adding stmts in notebook

* ML model improvement : also test the non-happy path

Co-authored-by: Gurunath LankupalliVenugopal <[email protected]>
Co-authored-by: Nils Braun <[email protected]>

Author: 3 people

dask_sql/context.py

@@ -105,6 +105,8 @@ def __init__(self):
         RelConverter.add_plugin_class(custom.ShowColumnsPlugin, replace=False)
         RelConverter.add_plugin_class(custom.ShowSchemasPlugin, replace=False)
         RelConverter.add_plugin_class(custom.ShowTablesPlugin, replace=False)
+        RelConverter.add_plugin_class(custom.ShowModelsPlugin, replace=False)
+        RelConverter.add_plugin_class(custom.ShowModelParamsPlugin, replace=False)
 
         RexConverter.add_plugin_class(core.RexCallPlugin, replace=False)
         RexConverter.add_plugin_class(core.RexInputRefPlugin, replace=False)

dask_sql/physical/rel/custom/__init__.py

@@ -3,10 +3,12 @@
 from .create_model import CreateModelPlugin
 from .create_table import CreateTablePlugin
 from .create_table_as import CreateTableAsPlugin
+from .describe_model import ShowModelParamsPlugin
 from .drop_model import DropModelPlugin
 from .drop_table import DropTablePlugin
 from .predict import PredictModelPlugin
 from .schemas import ShowSchemasPlugin
+from .show_models import ShowModelsPlugin
 from .tables import ShowTablesPlugin
 
 __all__ = [
@@ -20,4 +22,6 @@
     ShowColumnsPlugin,
     ShowSchemasPlugin,
     ShowTablesPlugin,
+    ShowModelsPlugin,
+    ShowModelParamsPlugin,
 ]

dask_sql/physical/rel/custom/describe_model.py

@@ -0,0 +1,33 @@
+import dask.dataframe as dd
+import pandas as pd
+
+from dask_sql.datacontainer import ColumnContainer, DataContainer
+from dask_sql.physical.rel.base import BaseRelPlugin
+
+
+class ShowModelParamsPlugin(BaseRelPlugin):
+    """
+    Show all Params used to train a given model along with the columns
+    used for training.
+    The SQL is:
+
+        DESCRIBE MODEL <model_name>
+
+    The result is also a table, although it is created on the fly.
+    """
+
+    class_name = "com.dask.sql.parser.SqlShowModelParams"
+
+    def convert(
+        self, sql: "org.apache.calcite.sql.SqlNode", context: "dask_sql.Context"
+    ) -> DataContainer:
+        model_name = str(sql.getModelName().getIdentifier())
+        if model_name not in context.models:
+            raise RuntimeError(f"A model with the name {model_name} is not present.")
+        model, training_columns = context.models[model_name]
+        model_params = model.get_params()
+        model_params["training_columns"] = training_columns.tolist()
+        df = pd.DataFrame.from_dict(model_params, orient="index", columns=["Params"])
+        cc = ColumnContainer(df.columns)
+        dc = DataContainer(dd.from_pandas(df, npartitions=1), cc)
+        return dc

dask_sql/physical/rel/custom/show_models.py

@@ -0,0 +1,28 @@
+import dask.dataframe as dd
+import pandas as pd
+
+from dask_sql.datacontainer import ColumnContainer, DataContainer
+from dask_sql.physical.rel.base import BaseRelPlugin
+
+
+class ShowModelsPlugin(BaseRelPlugin):
+    """
+    Show all MODELS currently registered/trained.
+    The SQL is:
+
+        SHOW MODELS
+
+    The result is also a table, although it is created on the fly.
+    """
+
+    class_name = "com.dask.sql.parser.SqlShowModels"
+
+    def convert(
+        self, sql: "org.apache.calcite.sql.SqlNode", context: "dask_sql.Context"
+    ) -> DataContainer:
+
+        df = pd.DataFrame({"Models": list(context.models.keys())})
+
+        cc = ColumnContainer(df.columns)
+        dc = DataContainer(dd.from_pandas(df, npartitions=1), cc)
+        return dc

dask_sql/utils.py

@@ -5,7 +5,7 @@
 from collections import defaultdict
 from contextlib import contextmanager
 from datetime import datetime
-from typing import Any, Dict, List
+from typing import Any, Dict, List, Tuple
 from unittest.mock import patch
 from uuid import uuid4
 

notebooks/Feature Overview.ipynb

@@ -285,6 +285,26 @@
     "SHOW COLUMNS FROM \"iris\""
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%sql\n",
+    "DESCRIBE iris"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%sql\n",
+    "DESCRIBE TABLE iris"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -507,6 +527,26 @@
     ")"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%sql\n",
+    "SHOW MODELS"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%sql\n",
+    "DESCRIBE MODEL my_model"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -571,13 +611,20 @@
     "\"\"\").compute() \n",
     "t.set_index([\"target\", \"species\"]).unstack(\"species\").number.plot.bar()"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {
   "kernelspec": {
    "display_name": "Python 3",
    "language": "python",
-   "name": "python3"
+   "name": "Python 3"
   },
   "language_info": {
    "codemirror_mode": {

planner/src/main/codegen/config.fmpp

@@ -36,6 +36,7 @@ data: {
         "com.dask.sql.parser.SqlShowColumns",
         "com.dask.sql.parser.SqlShowSchemas",
         "com.dask.sql.parser.SqlShowTables"
+        "com.dask.sql.parser.SqlShowModels"
       ]
 
       # List of keywords.
@@ -49,6 +50,7 @@ data: {
         "SCHEMAS"
         "STATISTICS"
         "TABLES"
+        "MODELS"
       ]
 
       # The keywords can only be used in a specific context,
@@ -68,11 +70,12 @@ data: {
       # List of methods for parsing custom SQL statements
       statementParserMethods: [
          "SqlAnalyzeTable()"
-         "SqlDescribeTable()"
          "SqlShowColumns()"
          "SqlShowSchemas()"
          "SqlShowTables()"
          "SqlPredictModel()"
+         "SqlShowModels()"
+         "SqlDescribeTableOrModel()"
       ]
 
       createStatementParserMethods: [

planner/src/main/codegen/includes/show.ftl

@@ -46,19 +46,31 @@ SqlNode SqlShowColumns() :
     }
 }
 
-// DESCRIBE "table"
-SqlNode SqlDescribeTable() :
+// DESCRIBE <table_name>
+// DESCRIBE TABLE <table_name>
+// DESCRIBE MODEL <model_name>
+SqlNode SqlDescribeTableOrModel() :
 {
     final Span s;
     final SqlIdentifier schemaName;
     final SqlIdentifier tableName;
+    final SqlModelIdentifier modelName;
 }
 {
     <DESCRIBE> { s = span(); }
-    tableName = CompoundTableIdentifier()
-    {
-        return new SqlShowColumns(s.end(this), tableName);
-    }
+    (
+        LOOKAHEAD(2)
+        modelName = ModelIdentifier()
+        {
+            return new SqlShowModelParams(s.end(this), modelName);
+        }
+    |
+        [ <TABLE> ]
+        tableName = CompoundTableIdentifier()
+        {
+            return new SqlShowColumns(s.end(this), tableName);
+        }
+    )
 }
 
 // ANALYZE TABLE table_identifier COMPUTE STATISTICS [ FOR COLUMNS col [ , ... ] | FOR ALL COLUMNS ]
@@ -86,3 +98,15 @@ SqlNode SqlAnalyzeTable() :
         return new SqlAnalyzeTable(s.end(this), tableName, columnList);
     }
 }
+
+// SHOW MODELS
+SqlNode SqlShowModels() :
+{
+   final Span s;
+}
+{
+    <SHOW> { s = span(); } <MODELS>
+    {
+        return new SqlShowModels(s.end(this));
+    }
+}

planner/src/main/java/com/dask/sql/parser/SqlShowModelParams.java

@@ -0,0 +1,40 @@
+package com.dask.sql.parser;
+import java.util.List;
+import org.apache.calcite.sql.SqlCall;
+import org.apache.calcite.sql.SqlIdentifier;
+import org.apache.calcite.sql.SqlNode;
+import org.apache.calcite.sql.SqlNodeList;
+import org.apache.calcite.sql.SqlOperator;
+import org.apache.calcite.sql.SqlWriter;
+import org.apache.calcite.sql.parser.SqlParserPos;
+
+
+public class SqlShowModelParams extends SqlCall {
+    final SqlModelIdentifier modelName;
+
+    public SqlShowModelParams(SqlParserPos pos, SqlModelIdentifier modelName) {
+        super(pos);
+        this.modelName = modelName;
+    }
+
+    @Override
+    public SqlOperator getOperator() {
+        throw new UnsupportedOperationException();
+    }
+
+    @Override
+    public List<SqlNode> getOperandList() {
+        throw new UnsupportedOperationException();
+    }
+
+    @Override
+    public void unparse(SqlWriter writer, int leftPrec, int rightPrec) {
+        writer.keyword("DESCRIBE");
+        writer.keyword("MODEL");
+        this.modelName.unparse(writer, leftPrec, rightPrec);
+    }
+    public SqlModelIdentifier getModelName() {
+        return this.modelName;
+    }
+
+}

planner/src/main/java/com/dask/sql/parser/SqlShowModels.java

@@ -0,0 +1,30 @@
+package com.dask.sql.parser;
+import java.util.ArrayList;
+import java.util.List;
+import org.apache.calcite.sql.SqlCall;
+import org.apache.calcite.sql.SqlOperator;
+import org.apache.calcite.sql.SqlWriter;
+import org.apache.calcite.sql.SqlNode;
+import org.apache.calcite.sql.parser.SqlParserPos;
+
+public class SqlShowModels extends SqlCall {
+
+    public SqlShowModels(SqlParserPos pos) {
+        super(pos);
+    }
+    public SqlOperator getOperator() {
+        throw new UnsupportedOperationException();
+    }
+
+    public List<SqlNode> getOperandList() {
+        ArrayList<SqlNode> operandList = new ArrayList<SqlNode>();
+        return operandList;
+    }
+
+    @Override
+    public void unparse(SqlWriter writer,int leftPrec, int rightPrec) {
+        writer.keyword("SHOW");
+        writer.keyword("MODELS");
+
+    }
+}