Model Technique

Model Technique is the algorithm/technique used to create python model using the library/package which was created in the Model Library Screen. It is the actual information captured in the MMG application that helps in training the model (Upload and Build).

1. In the LHS menu, click Model Catalog > Model Techniqueoption.
   The Technique Summarypage is displayed.

   Figure 8-38 Technique Summary page

   
   

   

   
   
2. In the Technique Summary page, click Add.
   The Add Technique page is displayed.

   Figure 8-39 Add Technique page

   
   

   

   
   
3. Enter the name of the technique.
4. Enter the description of the technique.
5. Select the library from the drop-down. Currently, MMG supports the libraries such as keras, ONNX, scikit-learn, and xgboost.
6. Select the type as either as Classification or Regression.
   Classification: Classification is a process of finding a function which helps in dividing the dataset into classes based on different parameters. The task of the classification algorithm is to find the mapping function to map the input(x) to the discrete output(y).
   Regression: Regression is a process of finding the correlations between dependent and independent variables. The task of the Regression algorithm is to find the mapping function to map the input variable(x) to the continuous output variable(y).
7. Enter the signatures such as Import, Load, Train, Save, and Infer. For more details on the signatures, click on the respective help icon.
   Some of the signatures captured in library stage might not be standard across different algorithm/technique provided by the library.
8. Navigate to Hyperparameter Details tab and click Add to add the parameters.
   You can add the type of parameters such as String, Integer, Float, and Boolean.
9. Click Create.
   The Model Technique is created and displayed in the Technique Summaryscreen.

   Note:

   MMG User needs to check the details provided above from the homepage of the library being captured.

   For example: https://scikit-learn.org/stable/.

   OR
   You can select the Use Template option to pre-fill the entries from the seeded list of Libraries.
   You can also edit the data based on your requirements.