Retrieve predictions that have previously been computed.
Training predictions encoded either as JSON or CSV.
If CSV output was requested, the returned CSV data will contain the following columns:
For regression projects: row_id and prediction.
For binary classification projects: row_id, prediction, class_<positive_class_label> and class_<negative_class_label>.
For multi classification projects: row_id, prediction and a class_<class_label> for each class.
For time-series, these additional columns will be added: forecast_point, forecast_distance, timestamp, and series_id.
.. minversion:: v2.21
* If `explanationAlgorithm` = 'shap', these additional columns will be added:
triplets of (`Explanation_<i>_feature_name`,
`Explanation_<i>_feature_value`, and `Explanation_<i>_strength`) for `i` ranging
from 1 to `maxExplanations`, `shap_remaining_total` and `shap_base_value`. Binary
classification projects will also have `explained_class`, the class for which
positive SHAP values imply an increased probability.
Click Try It! to start a request and see the response here!