Performance assessment

Endpoints to obtain fairness metrics for your AI models

Get performance metrics

GET https://api.expai.io/api/performance/<model_id>/get_metrics

For a given model in your projects, get the available performance metrics

Path Parameters

Name	Type	Description
model_id	string	Unique ID for the model

Headers

Name	Type	Description
access-token	string	Active JWT token

200 Successful execution. The values in the metrics dictionary should be used to identify them in later queries.

{
    "code": "success",
    "http_code": 200,
    "message": "success",
    "metrics": {
        "Mean absolute error (MAE)": "mean_absolute_error",
        "Mean squared error (MSE)": "mean_squared_error",
        "R2 Score": "r2_score",
        "Root mean squared error (RMSE)": "root_mean_squared_error"
    }
}

Compute Performance Metrics on subgroups

POST https://api.expai.io/api/performance/<model_id>/compute

This endpoint returns the desired performance metrics for your model on specific subset of data given by a variable.

Path Parameters

Name	Type	Description
model_id	string	Unique ID for the model you want to evaluate

Headers

Name	Type	Description
access-token	string	Active JWT token

Request Body

Name	Type	Description
metrics	array	List of metrics to be evaluated. You must provide the value of the dictionary returned by the previous endpoint.
sample_id	string	ID for the sample used to evaluate the model. It must contain a target column (supervised).
variable	string	Variable used to segment the data into groups.
variable_type	object	Dictionary of the shape {variable: <TYPE>}. Available types are: "numerical" and "categorical".
subset_indexes	array	Array containing the indexes of the filtered entries that should be considered.
subset_filters	object	Dictionary containing filtering rules for the dataframe.
target_class	string	If you are explaining a categorical model, choose the class for which you want to obtain insights.

200 Successful execution. Returns group metrics segmented by the desired variable and overall metrics for the whole dataset.

{
    "group_metrics": {
        "2012": {
            "Mean squared error (MSE)": 55597287.77921276,
            "R2 Score": 0.8464727420256901
        },
        "2013": {
            "Mean squared error (MSE)": 131866772.47007993,
            "R2 Score": 0.7984998271644645
        },
        "2014": {
            "Mean squared error (MSE)": 112957338.09533538,
            "R2 Score": 0.8934880347241547
        },
        "2015": {
            "Mean squared error (MSE)": 164460968.7939401,
            "R2 Score": 0.8908172129393476
        },
        "2016": {
            "Mean squared error (MSE)": 260609158.18212545,
            "R2 Score": 0.9406037510358035
        },
    "http_code": 200,
    "message": "success",
    "overall_metrics": {
        "Mean squared error (MSE)": 45145913.93682745,
        "R2 Score": 0.9235097053991715
    }
}