Run Clustering Analysis with APPLY Table Operator | K-Means Clustering Classification| Open Analytics Framework - Running Clustering Analysis with APPLY Table Operator

Create a teradataml Apply class object with the characteristics you want to consider for the call to the APPLY Table Operator.

In this example, specify the following:

The data argument so you can specify the input teradataml DataFrame that points to the input data table.
The apply_command argument to call the Python 3 interpreter in your user environment and run your script.
Assume you would like to determine 7 clusters for each observation group in your input data. The Python command specification, then, must be accompanied by an additional argument with the value 7 to follow the script filename.
The data_partition_column argument with the value 'ObsGroup' to specify the input data be partitioned and analyzed separately according to their observation group value.
The data_order_column argument with the value 'ObsID' to specify that rows in each partition be ordered by the ID value of each observation.
The returns argument with the list of output variables and types returned by your script.
The env_name argument to specify your user environment handler.

Assume your Python script returns scored rows that contain the following variables:

A string variable ObsID.
A string variable ObsGrp to designate the group an observation belongs to.
A string variable ClustID to designate the cluster in its observation group that an observation has been classified into by the K-Means algorithm.
A set of string variables X_Centroid and Y_Centroid to designate the centroid coordinates of the cluster the observation belongs to.
A set of string variables ObsSilhCoeff and SilhScore that report the clustering analysis Silhouette coefficient and score, respectively, for the current observation.

Call to the Apply class.

apply_obj = Apply(data = clustData,
                  apply_command = 'python3 clustering.py 7',
                  data_partition_column = 'ObsGroup',
                  data_order_column = 'ObsID',
                  returns = {"ObsID": VARCHAR(20), "ObsGrp": VARCHAR(20),
                             "ClustID": VARCHAR(10),
                             "X_Centroid": VARCHAR(30),
                             "Y_Centroid": VARCHAR(30),
                             "Dummy1": VARCHAR(30),
                             "ObsSilhCoeff": VARCHAR(30), 
                             "SilhScore": VARCHAR(30)},
                  env_name = demo_env
                 )

You can print on screen the SQL query submitted by teradataml to VantageCloud Lake with the following statement:

display.print_sqlmr_query = True

Run the Python script inside the user environment with the execute_script method of the Apply class object.

Observe that after running the Python statement, the system prints for you the corresponding SQL query as requested before producing the results.

apply_obj.execute_script().head(5)

SELECT * FROM Apply(
        ON “VCLUSER”."clustData" AS "input"
        PARTITION BY ObsGroup
        ORDER BY "ObsID" ASC NULLS FIRST
        returns(ObsID VARCHAR(20), ObsGrp VARCHAR(20), ClustID VARCHAR(10),
                X_Centroid VARCHAR(30), Y_Centroid VARCHAR(30),
                Dummy1 VARCHAR(30), ObsSilhCoeff VARCHAR(30),
                SilhScore VARCHAR(30))
        USING
        APPLY_COMMAND('python3 clustering.py 7')
        ENVIRONMENT('opaf_my_clust_env')
        STYLE('csv')
        delimiter(',')
) as sqlmr

Out:

        ObsID  ObsGrp ClustID        X_Centroid                Y_Centroid            Dummy1      ObsSilhCoeff           SilhScore
0        1           1        3        0.7417645912751678        0.8532491845637584        7        0.048249686795930        0.3703761282659183
1        7910        9        1        0.5008595791666667        0.14966305000000002        7        0.630966145110239        0.3707661187689244
2        3027        4        2        0.5332838275862068        0.2166037931034483        7        0.437218587802698        0.3687863623349997
3        5034        6        4        0.8407428916666666        0.7530141347222221        7        0.239935608340830        0.3842943593843244
4        2           1        1        0.5070352009421265        0.5366947352624495        7        0.610658798723727        0.3703761282659183

Run Clustering Analysis with APPLY Table Operator | K-Means Clustering Classification| Open Analytics Framework - Running Clustering Analysis with APPLY Table Operator - Teradata Vantage

Teradata® VantageCloud Lake