get_metric {civis}R Documentation

Get model metrics

Description

Get model metrics

Usage

get_metric(model, name = NULL)

Arguments

model

Object from civis_ml

name

Name of the metric. If NULL, all metrics are returned. Possible metrics are listed in details.

Details

The list of possible metrics provided by CivisML is:

Classification:

"accuracy", "confusion_matrix", "p_correct", "pop_incidence_true", "pop_incidence_pred", "roc_auc", "log_loss", "brier_score", "roc_curve", "calibration_curve", "deciles", "score_histogram", "training_histogram", "oos_score_table".

Regression:

"mad", "rmse", "r_squared", "score_histogram", "training_histogram", "y_yhat_plot", "y_yhat_outlier_rows", "oos_score_table"

Value

The metric given by name, or all metrics.


[Package civis version 3.0.0 Index]