teradataml 17.20.0.6__py3-none-any.whl → 20.0.0.0__py3-none-any.whl

teradataml/analytics/mle/ROC.py

@@ -1,425 +0,0 @@
-#!/usr/bin/python
-# ################################################################## 
-# 
-# Copyright 2018 Teradata. All rights reserved.
-# TERADATA CONFIDENTIAL AND TRADE SECRET
-# 
-# Primary Owner: Rohit Agrawal (rohit.agrawal@teradata.com)
-# Secondary Owner: Pankaj Purandare (pankajvinod.purandare@teradata.com)
-# 
-# Version: 1.2
-# Function Version: 1.8
-# 
-# ################################################################## 
-
-import inspect
-import time
-from teradataml.common.wrapper_utils import AnalyticsWrapperUtils
-from teradataml.common.utils import UtilFuncs
-from teradataml.context.context import *
-from teradataml.dataframe.dataframe import DataFrame
-from teradataml.common.aed_utils import AedUtils
-from teradataml.analytics.analytic_query_generator import AnalyticQueryGenerator
-from teradataml.common.exceptions import TeradataMlException
-from teradataml.common.messages import Messages
-from teradataml.common.messagecodes import MessageCodes
-from teradataml.common.constants import TeradataConstants
-from teradataml.dataframe.dataframe_utils import DataFrameUtils as df_utils
-from teradataml.options.display import display
-
-class ROC:
-    
-    def __init__(self,
-        data = None,
-        model_id_column = None,
-        probability_column = None,
-        observation_column = None,
-        positive_class = None,
-        num_thresholds = 50,
-        data_sequence_column = None):
-        """
-        DESCRIPTION:
-            A receiver operating characteristic (ROC) curve shows the performance 
-            of a binary classification model as its discrimination threshold 
-            varies. For a range of thresholds, the curve plots the true positive 
-            rate against the false positive rate.
-            Note:
-                This function is available only when teradataml is connected to
-                Vantage 1.1 or later versions.
-
-         
-        PARAMETERS:
-            data:
-                Required Argument.
-                Specifies a teradataml DataFrame that contains the
-                prediction-actual pairs for a binary classifier.
-            
-            model_id_column:
-                Optional Argument.
-                Specifies the input teradataml DataFrame column that
-                contains the model or partition identifiers for the ROC curves.
-                Use this argument only when input teradataml DataFrame contains
-                information for more than one model. The function creates a separate
-                ROC curve for each model identifier in this column. Each model must
-                include exactly two classes in observation_column.
-                Types: str
-            
-            probability_column:
-                Required Argument.
-                Specifies the input teradataml DataFrame column that
-                contains the predictions.
-                Types: str
-            
-            observation_column:
-                Required Argument.
-                Specifies the input teradataml DataFrame column that
-                contains the actual classes. 
-                Types: str
-            
-            positive_class:
-                Required Argument.
-                Specifies the label of the positive class.
-                Types: str
-            
-            num_thresholds:
-                Optional Argument.
-                Specifies the number of thresholds for the function to use. The 
-                num_threshold must be a Integer value in the range [1, 10000]. The
-                function uniformly distributes the thresholds between 0 and 1.
-                Default Value: 50
-                Types: int
-            
-            data_sequence_column:
-                Optional Argument.
-                Specifies the list of column(s) that uniquely identifies each row of 
-                the input argument "data". The argument is used to ensure 
-                deterministic results for functions which produce results that vary 
-                from run to run.
-                Types: str OR list of Strings (str)
-         
-        RETURNS:
-            Instance of ROC.
-            Output teradataml DataFrames can be accessed using attribute 
-            references, such as ROCObj.<attribute_name>.
-            Output teradataml DataFrame attribute names are:
-                1. roc_output
-                2. output
-         
-        Note:
-            1. Function will return auc and gini values in output teradataml DataFrame.
-            2. Function will return roc values (thresholds, false positive rates, and
-               true positive rates) in roc_output teradataml DataFrame.
-         
-         
-        RAISES:
-            TeradataMlException
-         
-         
-        EXAMPLES:
-            # Load the data to run the example.
-            load_example_data("ROC", "roc_input")
-         
-            # Create teradataml DataFrame.
-            roc_input = DataFrame.from_table("roc_input")
-         
-            # Example : Running ROC function with default values.
-            # It will return the result DataFrame roc_output and output.
-            roc_out1 = ROC(data=roc_input,
-                          probability_column='probability',
-                          observation_column='observation',
-                          model_id_column='model_id',
-                          positive_class='1',
-                          num_thresholds=100
-                          )
-         
-            # Print the result DataFrame.
-            print(roc_out1.roc_output)
-            print(roc_out1.output)
-        
-        """
-        
-        # Start the timer to get the build time
-        _start_time = time.time()
-        
-        self.data  = data 
-        self.model_id_column  = model_id_column 
-        self.probability_column  = probability_column 
-        self.observation_column  = observation_column 
-        self.positive_class  = positive_class 
-        self.num_thresholds  = num_thresholds
-        self.data_sequence_column  = data_sequence_column 
-        
-        # Create TeradataPyWrapperUtils instance which contains validation functions.
-        self.__awu = AnalyticsWrapperUtils()
-        self.__aed_utils = AedUtils()
-        
-        # Create argument information matrix to do parameter checking
-        self.__arg_info_matrix = []
-        self.__arg_info_matrix.append(["data", self.data, False, (DataFrame)])
-        self.__arg_info_matrix.append(["model_id_column", self.model_id_column, True, (str)])
-        self.__arg_info_matrix.append(["probability_column", self.probability_column, False, (str)])
-        self.__arg_info_matrix.append(["observation_column", self.observation_column, False, (str)])
-        self.__arg_info_matrix.append(["positive_class", self.positive_class, False, (str)])
-        self.__arg_info_matrix.append(["num_thresholds", self.num_thresholds, True, (int)])
-        self.__arg_info_matrix.append(["data_sequence_column", self.data_sequence_column, True, (str,list)])
-        
-        if inspect.stack()[1][3] != '_from_model_catalog':
-            # Perform the function validations
-            self.__validate()
-            # Generate the ML query
-            self.__form_tdml_query()
-            # Execute ML query
-            self.__execute()
-            # Get the prediction type
-            self._prediction_type = self.__awu._get_function_prediction_type(self)
-        
-        # End the timer to get the build time
-        _end_time = time.time()
-        
-        # Calculate the build time
-        self._build_time = (int)(_end_time - _start_time)
-        
-    def __validate(self):
-        """
-        Function to validate sqlmr function arguments, which verifies missing 
-        arguments, input argument and table types. Also processes the 
-        argument values.
-        """
-        
-        # Make sure that a non-NULL value has been supplied for all mandatory arguments
-        self.__awu._validate_missing_required_arguments(self.__arg_info_matrix)
-        
-        # Make sure that a non-NULL value has been supplied correct type of argument
-        self.__awu._validate_argument_types(self.__arg_info_matrix)
-        
-        # Check to make sure input table types are strings or data frame objects or of valid type.
-        self.__awu._validate_input_table_datatype(self.data, "data", None)
-        
-        # Check whether the input columns passed to the argument are not empty.
-        # Also check whether the input columns passed to the argument valid or not.
-        self.__awu._validate_input_columns_not_empty(self.probability_column, "probability_column")
-        self.__awu._validate_dataframe_has_argument_columns(self.probability_column, "probability_column", self.data, "data", False)
-        
-        self.__awu._validate_input_columns_not_empty(self.observation_column, "observation_column")
-        self.__awu._validate_dataframe_has_argument_columns(self.observation_column, "observation_column", self.data, "data", False)
-        
-        self.__awu._validate_input_columns_not_empty(self.model_id_column, "model_id_column")
-        self.__awu._validate_dataframe_has_argument_columns(self.model_id_column, "model_id_column", self.data, "data", False)
-        
-        self.__awu._validate_input_columns_not_empty(self.data_sequence_column, "data_sequence_column")
-        self.__awu._validate_dataframe_has_argument_columns(self.data_sequence_column, "data_sequence_column", self.data, "data", False)
-        
-        
-    def __form_tdml_query(self):
-        """
-        Function to generate the analytical function queries. The function defines 
-        variables and list of arguments required to form the query.
-        """
-        # Generate temp table names for output table parameters if any.
-        self.__roc_output_temp_tablename = UtilFuncs._generate_temp_table_name(prefix="td_roc0", use_default_database=True, gc_on_quit=True, quote=False, table_type=TeradataConstants.TERADATA_TABLE)
-        
-        # Output table arguments list
-        self.__func_output_args_sql_names = ["ROCTable"]
-        self.__func_output_args = [self.__roc_output_temp_tablename]
-        
-        # Model Cataloging related attributes.
-        self._sql_specific_attributes = {}
-        self._sql_formula_attribute_mapper = {}
-        self._target_column = None
-        self._algorithm_name = None
-        
-        # Generate lists for rest of the function arguments
-        self.__func_other_arg_sql_names = []
-        self.__func_other_args = []
-        self.__func_other_arg_json_datatypes = []
-        
-        self.__func_other_arg_sql_names.append("ProbabilityColumn")
-        self.__func_other_args.append(UtilFuncs._teradata_collapse_arglist(UtilFuncs._teradata_quote_arg(self.probability_column, "\""), "'"))
-        self.__func_other_arg_json_datatypes.append("COLUMN_NAMES")
-        
-        self.__func_other_arg_sql_names.append("ObservationColumn")
-        self.__func_other_args.append(UtilFuncs._teradata_collapse_arglist(UtilFuncs._teradata_quote_arg(self.observation_column, "\""), "'"))
-        self.__func_other_arg_json_datatypes.append("COLUMN_NAMES")
-        
-        if self.model_id_column is not None:
-            self.__func_other_arg_sql_names.append("ModelIdColumn")
-            self.__func_other_args.append(UtilFuncs._teradata_collapse_arglist(UtilFuncs._teradata_quote_arg(self.model_id_column, "\""), "'"))
-            self.__func_other_arg_json_datatypes.append("COLUMN_NAMES")
-        
-        self.__func_other_arg_sql_names.append("PositiveClass")
-        self.__func_other_args.append(UtilFuncs._teradata_collapse_arglist(self.positive_class, "'"))
-        self.__func_other_arg_json_datatypes.append("STRING")
-        
-        if self.num_thresholds is not None and self.num_thresholds != 50:
-            self.__func_other_arg_sql_names.append("NumThresholds")
-            self.__func_other_args.append(UtilFuncs._teradata_collapse_arglist(self.num_thresholds, "'"))
-            self.__func_other_arg_json_datatypes.append("INTEGER")
-        
-        # Generate lists for rest of the function arguments
-        sequence_input_by_list = []
-        if self.data_sequence_column is not None:
-            sequence_input_by_list.append("InputTable:" + UtilFuncs._teradata_collapse_arglist(self.data_sequence_column, ""))
-        
-        if len(sequence_input_by_list) > 0:
-            self.__func_other_arg_sql_names.append("SequenceInputBy")
-            sequence_input_by_arg_value = UtilFuncs._teradata_collapse_arglist(sequence_input_by_list, "'")
-            self.__func_other_args.append(sequence_input_by_arg_value)
-            self.__func_other_arg_json_datatypes.append("STRING")
-            self._sql_specific_attributes["SequenceInputBy"] = sequence_input_by_arg_value
-        
-        
-        # Declare empty lists to hold input table information.
-        self.__func_input_arg_sql_names = []
-        self.__func_input_table_view_query = []
-        self.__func_input_dataframe_type = []
-        self.__func_input_distribution = []
-        self.__func_input_partition_by_cols = []
-        self.__func_input_order_by_cols = []
-        
-        # Process data
-        self.__table_ref = self.__awu._teradata_on_clause_from_dataframe(self.data, False)
-        self.__func_input_distribution.append("NONE")
-        self.__func_input_arg_sql_names.append("InputTable")
-        self.__func_input_table_view_query.append(self.__table_ref["ref"])
-        self.__func_input_dataframe_type.append(self.__table_ref["ref_type"])
-        self.__func_input_partition_by_cols.append("NA_character_")
-        self.__func_input_order_by_cols.append("NA_character_")
-        
-        function_name = "ROC"
-        # Create instance to generate SQLMR.
-        self.__aqg_obj = AnalyticQueryGenerator(function_name, 
-                self.__func_input_arg_sql_names, 
-                self.__func_input_table_view_query, 
-                self.__func_input_dataframe_type, 
-                self.__func_input_distribution, 
-                self.__func_input_partition_by_cols, 
-                self.__func_input_order_by_cols, 
-                self.__func_other_arg_sql_names, 
-                self.__func_other_args, 
-                self.__func_other_arg_json_datatypes, 
-                self.__func_output_args_sql_names, 
-                self.__func_output_args, 
-                engine="ENGINE_ML")
-        # Invoke call to SQL-MR generation.
-        self.sqlmr_query = self.__aqg_obj._gen_sqlmr_select_stmt_sql()
-        
-        # Print SQL-MR query if requested to do so.
-        if display.print_sqlmr_query:
-            print(self.sqlmr_query)
-        
-        # Set the algorithm name for Model Cataloging.
-        self._algorithm_name = self.__aqg_obj._get_alias_name_for_function(function_name)
-        
-    def __execute(self):
-        """
-        Function to execute SQL-MR queries. 
-        Create DataFrames for the required SQL-MR outputs.
-        """
-        # Generate STDOUT table name and add it to the output table list.
-        sqlmr_stdout_temp_tablename = UtilFuncs._generate_temp_table_name(prefix="td_sqlmr_out_", use_default_database=True, gc_on_quit=True, quote=False, table_type=TeradataConstants.TERADATA_TABLE)
-        try:
-            # Generate the output.
-            UtilFuncs._create_table(sqlmr_stdout_temp_tablename, self.sqlmr_query)
-        except Exception as emsg:
-            raise TeradataMlException(Messages.get_message(MessageCodes.TDMLDF_EXEC_SQL_FAILED, str(emsg)), MessageCodes.TDMLDF_EXEC_SQL_FAILED)
-        
-        # Update output table data frames.
-        self._mlresults = []
-        self.roc_output = self.__awu._create_data_set_object(df_input=UtilFuncs._extract_table_name(self.__roc_output_temp_tablename), source_type="table", database_name=UtilFuncs._extract_db_name(self.__roc_output_temp_tablename))
-        self.output = self.__awu._create_data_set_object(df_input=UtilFuncs._extract_table_name(sqlmr_stdout_temp_tablename), source_type="table", database_name=UtilFuncs._extract_db_name(sqlmr_stdout_temp_tablename))
-        self._mlresults.append(self.roc_output)
-        self._mlresults.append(self.output)
-        
-    def show_query(self):
-        """
-        Function to return the underlying SQL query.
-        When model object is created using retrieve_model(), then None is returned.
-        """
-        return self.sqlmr_query
-        
-    def get_prediction_type(self):
-        """
-        Function to return the Prediction type of the algorithm.
-        When model object is created using retrieve_model(), then the value returned is 
-        as saved in the Model Catalog.
-        """
-        return self._prediction_type
-        
-    def get_target_column(self):
-        """
-        Function to return the Target Column of the algorithm.
-        When model object is created using retrieve_model(), then the value returned is 
-        as saved in the Model Catalog.
-        """
-        return self._target_column
-        
-    def get_build_time(self):
-        """
-        Function to return the build time of the algorithm in seconds.
-        When model object is created using retrieve_model(), then the value returned is 
-        as saved in the Model Catalog.
-        """
-        return self._build_time
-        
-    def _get_algorithm_name(self):
-        """
-        Function to return the name of the algorithm.
-        """
-        return self._algorithm_name
-        
-    def _get_sql_specific_attributes(self):
-        """
-        Function to return the dictionary containing the SQL specific attributes of the algorithm.
-        """
-        return self._sql_specific_attributes
-        
-    @classmethod
-    def _from_model_catalog(cls,
-        roc_output = None,
-        output = None,
-        **kwargs):
-        """
-        Classmethod is used by Model Cataloging, to instantiate this wrapper class.
-        """
-        kwargs.pop("roc_output", None)
-        kwargs.pop("output", None)
-        
-        # Model Cataloging related attributes.
-        target_column = kwargs.pop("__target_column", None)
-        prediction_type = kwargs.pop("__prediction_type", None)
-        algorithm_name = kwargs.pop("__algorithm_name", None)
-        build_time = kwargs.pop("__build_time", None)
-        
-        # Let's create an object of this class.
-        obj = cls(**kwargs)
-        obj.roc_output  = roc_output 
-        obj.output  = output 
-        
-        # Initialize the sqlmr_query class attribute.
-        obj.sqlmr_query = None
-        
-        # Initialize the SQL specific Model Cataloging attributes.
-        obj._sql_specific_attributes = None
-        obj._target_column = target_column
-        obj._prediction_type = prediction_type
-        obj._algorithm_name = algorithm_name
-        obj._build_time = build_time
-        
-        # Update output table data frames.
-        obj._mlresults = []
-        obj.roc_output = obj.__awu._create_data_set_object(df_input=UtilFuncs._extract_table_name(obj.roc_output), source_type="table", database_name=UtilFuncs._extract_db_name(obj.roc_output))
-        obj.output = obj.__awu._create_data_set_object(df_input=UtilFuncs._extract_table_name(obj.output), source_type="table", database_name=UtilFuncs._extract_db_name(obj.output))
-        obj._mlresults.append(obj.roc_output)
-        obj._mlresults.append(obj.output)
-        return obj
-        
-    def __repr__(self):
-        """
-        Returns the string representation for a ROC class instance.
-        """
-        repr_string="############ STDOUT Output ############"
-        repr_string = "{}\n\n{}".format(repr_string,self.output)
-        repr_string="{}\n\n\n############ roc_output Output ############".format(repr_string)
-        repr_string = "{}\n\n{}".format(repr_string,self.roc_output)
-        return repr_string
-