CLI commands#

Module for command-line interface (CLI).

Basic `forecast` usage:#

Usage: etna forecast [OPTIONS] CONFIG_PATH TARGET_PATH FREQ OUTPUT_PATH [EXOG_PATH]
                     [FORECAST_CONFIG_PATH] [RAW_OUTPUT] [KNOWN_FUTURE]

Command to make forecast with etna without coding.

Arguments:
    CONFIG_PATH             path to yaml config with desired pipeline  [required]
    TARGET_PATH             path to csv with data to forecast  [required]
    FREQ                    frequency of timestamp in files in pandas format  [required]
    OUTPUT_PATH             where to save forecast  [required]
    [EXOG_PATH]             path to csv with exog data
    [FORECAST_CONFIG_PATH]  path to yaml config with forecast params
    [RAW_OUTPUT]            by default we return only forecast without features [default: False]
    [KNOWN_FUTURE]          list of all known_future columns (regressor columns). If not specified then all exog_columns considered known_future [default: None]

Forecast config parameters

start_timestamp - timestamp with the starting point of forecast.
estimate_n_folds - whether to estimate the number of folds from data. Works only when prediction intervals are enabled. Requires context_size parameter set in pipeline config.

Other parameters that could be set in the configuration file could be found in forecast() method documentation.

Setting these parameters is optional.

Pipeline config parameters

context_size - minimum number of points in the history that is required by pipeline to produce a forecast.

Further information on pipeline parameters could be found in Pipeline method documentation.

How to create config?

Example of pipeline’s config:

_target_: etna.pipeline.Pipeline
horizon: 4
model:
  _target_: etna.models.CatBoostMultiSegmentModel
transforms:
  - _target_: etna.transforms.LinearTrendTransform
    in_column: target
  - _target_: etna.transforms.SegmentEncoderTransform

Example of forecast params config:

prediction_interval: true
quantiles: [0.025, 0.975]
n_folds: 3

Parameter start_timestamp could be set similarly:

prediction_interval: true
quantiles: [0.025, 0.975]
start_timestamp: "2020-01-12"

Example of a pair of configs for number of folds estimation:

_target_: etna.pipeline.Pipeline
horizon: 4
model:
  _target_: etna.models.CatBoostMultiSegmentModel
transforms:
  - _target_: etna.transforms.LinearTrendTransform
    in_column: target
  - _target_: etna.transforms.SegmentEncoderTransform
context_size: 1

prediction_interval: true
quantiles: [0.025, 0.975]
estimate_n_folds: true

How to prepare data?

Example of dataset with data to forecast:

timestamp	segment	target
2020-01-01	segment_1	1
2020-01-02	segment_1	2
2020-01-03	segment_1	3
2020-01-04	segment_1	4
…
2020-01-10	segment_2	10
2020-01-11	segment_2	20

Example of exog dataset:

timestamp	segment	regressor_1	regressor_2
2020-01-01	segment_1	11	12
2020-01-02	segment_1	22	13
2020-01-03	segment_1	31	14
2020-01-04	segment_1	42	15
…
2020-02-10	segment_2	101	61
2020-02-11	segment_2	205	54

Basic `backtest` usage:#

Usage: etna backtest [OPTIONS] CONFIG_PATH BACKTEST_CONFIG_PATH TARGET_PATH FREQ OUTPUT_PATH [EXOG_PATH] [KNOWN_FUTURE]

Command to run backtest with etna without coding.

Arguments:
    CONFIG_PATH             path to yaml config with desired pipeline  [required]
    BACKTEST_CONFIG_PATH    path to yaml with backtest run config [required]
    TARGET_PATH             path to csv with data to forecast  [required]
    FREQ                    frequency of timestamp in files in pandas format  [required]
    OUTPUT_PATH             where to save forecast  [required]
    [EXOG_PATH]             path to csv with exog data
    [KNOWN_FUTURE]          list of all known_future columns (regressor columns). If not specified then all exog_columns considered known_future [default: None]

Backtest config parameters

estimate_n_folds - whether to estimate the number of folds from data. Requires context_size parameter set in pipeline config.

Other parameters that could be set in the configuration file could be found in backtest() method documentation.

Setting these parameters is optional.

Pipeline config parameters

context_size - minimum number of points in the history that is required by pipeline to produce a forecast.

Further information on pipeline parameters could be found in Pipeline method documentation.

How to create configs?

Example of pipeline’s config:

_target_: etna.pipeline.Pipeline
horizon: 4
model:
  _target_: etna.models.CatBoostMultiSegmentModel
transforms:
  - _target_: etna.transforms.LinearTrendTransform
    in_column: target
  - _target_: etna.transforms.SegmentEncoderTransform

Example of backtest run config:

n_folds: 3
n_jobs: 3
metrics:
  - _target_: etna.metrics.MAE
  - _target_: etna.metrics.MSE
  - _target_: etna.metrics.MAPE
  - _target_: etna.metrics.SMAPE

Example of a pair of configs for number of folds estimation for backtest:

_target_: etna.pipeline.Pipeline
horizon: 4
model:
  _target_: etna.models.CatBoostMultiSegmentModel
transforms:
  - _target_: etna.transforms.LinearTrendTransform
    in_column: target
  - _target_: etna.transforms.SegmentEncoderTransform
context_size: 1

n_folds: 200
n_jobs: 4
metrics:
  - _target_: etna.metrics.MAE
  - _target_: etna.metrics.SMAPE
estimate_n_folds: true