Inferencing

The 'Inferencing' feature in AryaXAI allows users to apply a trained model to make predictions on specific files or tags. Here, you can execute predictions using the activated model on specific files or tags. You can run the model on Tag or files directly on the batch model. 

You can download these files as csv too.

The below function performs predictions on testing data using the XGBoost default model. You have the flexibility to pass any model you prefer or leave it blank to use the default model.

# model inference
project.model_inference(tag="Training",model_name="XGBoost_default")

# model_name optional default to active model for the project

The inferencing results are stored as 'Testing_XGBoost_v1_Inference' tag.

Help function on model inferencing:

# Help on method model_inference
help(project.model_inference)

To get a quick overview of testresults data:

testresults.head()

This result data will have additional columns named:

• "Predicted_value_AutoML": For classification model, value will be class of the prediction post mapping. For Ex: if the target has 3 values, then we map these values with 1,2,3. For regression, both predicted value and category remains same.
• "Prediction_category_AutoML": For classification model, here you can see the actual category of the target.
• "pred_proba_AutoML": This is the probability of the model.

Which are appended to the same data.

Retrieve the tags that are created by both user and the system:

#To retrieve the tags that are created by both user and the system
project.all_tags()

#To retrieve the tags that are created by user
project.tags()

‍

XAI Evals

The XAI Evals fetches feature values and their corresponding attribution scores for tabular ML models, enabling detailed evaluation and interpretability of model predictions. Use this method to gain insights into the model's decision-making process by visualizing how individual features influence predictions on the selected dataset.

Evals method:

project.evals_tabular(model_name="XGBoost_default", tag="training")

‍