Predict Command - Make predictions using trained models via CLI.
Functions
parse_extra_args(args)
Parse unrecognised command-line arguments into a keyword dictionary.
Leading dashes are stripped from each flag name. A flag followed by a non-flag token takes that token as its value, coerced to
bool, int, or float when possible; a flag with no value becomes True.Parameters
args
list of str
Raw leftover tokens, typically
click.Context.args.
Returns
kwargs
dict
Mapping of flag name to parsed value.
tuiml predict [OPTIONS]
Make predictions with a trained model.
Loads a model by ID or file path, applies it to a dataset, and prints a summary of the predictions, optionally writing them to CSV. Use
--stage to ask for class probabilities instead of labels, or to forecast future steps with a timeseries model. Anomaly detectors additionally report how many instances were flagged.Predict with a saved model and write the results to CSV:
bash
$ tuiml predict --model-path model.pkl -d new_data.csv -o predictions.csvPredict using the model ID printed by "tuiml train":
bash
$ tuiml predict --model-id abc123 -d new_data.csvGet class probabilities instead of labels:
bash
$ tuiml predict --model-path model.pkl -d new_data.csv --stage predict_probaForecast the next 12 steps of a timeseries model:
bash
$ tuiml predict --model-path arima.pkl --stage forecast --steps 12Options
--model-path
Path to a saved model file (alternative to --model-id).
--data, -d
Path to a data file or a built-in dataset name to predict on.
--model-id
Model ID printed by "tuiml train" (alternative to --model-path).
--steps
int
Number of steps to forecast ahead. Timeseries models only.
--output, -o
Write the predictions to this file as CSV.
--stage
Prediction stage to run: 'predict', 'predict_proba', or 'forecast'. Defaults to 'predict'.
--json-output
flag
Print the raw JSON result instead of the formatted summary.
--verbose, -v
flag
Echo the resolved configuration and re-raise full tracebacks on error.