QuickstartΒΆ

  1. Install the package

(venv) $ pip install xailens

2 .Add at the end of your model code:

import xailens

# define the dataschema
data_schema = xailens.DataSchema(
    # name of the unique identifier column for your instances
    instance_id_column="instance_uuid",
    # name of your target column, from y_test
    target_column="target",
    # name of your prediction column, from y_pred
    prediction_column="pred_value")

# your data and predictions
model_data = xailens.ModelData(
    data_schema=data_schema,        # DataSchema from above
    raw_instances=raw,              # your raw data
    x_test=X_test,                  # your X_test dataframe
    y_pred=y_pred,                  # your y_pred dataframe
    y_test=y_test)                  # your y_test dataframe

# overall context
ctx = xailens.ModelContext(
    # model type, can be "logistic_regression"," xgboost" or "llm"
    "logistic_regression",
    # XAI methods to apply, options are "coefficients", "tree_shap",
    # "llm_reasoning"
    ["coefficients"],
    # identifier for your dataset
    "cleveland-heart-data",
    # ModelData from above
    model_data,
    # your model object (that will be saved as joblib file)
    model=model,
    # directory name to save to (can be blank)
    model_dir_name="heart_lr",
    # display name for your model, as it will appear in the dashboard
    model_display_name="Logistic Regression")

xailens.run(ctx)

  1. Start the dashboard

(venv) $ xailens-dash