mirror of
https://github.com/OpenBB-finance/OpenBB.git
synced 2026-10-28 21:12:16 +00:00
e1bba247e2
* init commit for doc string fixes * quick fix * Forecast Guide Draft 1 * grammar * sentence edit * missing word * target column is a string, not a list. ooops * clarification on default target column * minor mods * more fixes * Fixed dashboard * val instead of tickerseries * Added fixes * $ --> <> * Fixed pylint * one quick change Co-authored-by: Danglewood <85772166+deeleeramone@users.noreply.github.com> Co-authored-by: Colin Delahunty <72827203+colin99d@users.noreply.github.com> Co-authored-by: colin99d <colin99delahunty@gmail.com>
105 lines
2.8 KiB
Python
105 lines
2.8 KiB
Python
# pylint: disable=too-many-arguments
|
|
"""Regression Model"""
|
|
__docformat__ = "numpy"
|
|
|
|
import logging
|
|
from typing import Tuple, Union, List
|
|
import warnings
|
|
|
|
|
|
import pandas as pd
|
|
|
|
from darts import TimeSeries
|
|
from darts.models import RegressionModel
|
|
from openbb_terminal.decorators import log_start_end
|
|
|
|
from openbb_terminal.forecast import helpers
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_regression_data(
|
|
data: Union[pd.Series, pd.DataFrame],
|
|
target_column: str = "close",
|
|
n_predict: int = 5,
|
|
past_covariates: str = None,
|
|
train_split: float = 0.85,
|
|
forecast_horizon: int = 5,
|
|
output_chunk_length: int = 5,
|
|
lags: Union[int, List[int]] = 14,
|
|
) -> Tuple[
|
|
List[TimeSeries], List[TimeSeries], List[TimeSeries], float, type[RegressionModel]
|
|
]:
|
|
"""Perform Regression Forecasting
|
|
|
|
Parameters
|
|
----------
|
|
data: Union[pd.Series, pd.DataFrame]
|
|
Input Data
|
|
n_predict: int
|
|
Days to predict. Defaults to 5.
|
|
target_column: str
|
|
Target column to forecast. Defaults to "close".
|
|
train_split: float
|
|
Train/val split. Defaults to 0.85.
|
|
past_covariates: str
|
|
Multiple secondary columns to factor in when forecasting. Defaults to None.
|
|
forecast_horizon: int
|
|
Forecast horizon when performing historical forecasting. Defaults to 5.
|
|
output_chunk_length: int
|
|
The length of the forecast of the model. Defaults to 1.
|
|
lags: Union[int, List[int]]
|
|
lagged target values to predict the next time step
|
|
|
|
Returns
|
|
-------
|
|
Tuple[List[TimeSeries], List[TimeSeries], List[TimeSeries], float, type[RegressionModel]]
|
|
Adjusted Data series,
|
|
Historical forecast by best RNN model,
|
|
list of Predictions,
|
|
Mean average precision error,
|
|
Best Regression Model.
|
|
"""
|
|
|
|
use_scalers = False
|
|
probabilistic = False
|
|
|
|
scaler, ticker_series = helpers.get_series(
|
|
data, target_column, is_scaler=use_scalers
|
|
)
|
|
|
|
past_covariate_whole, _, _ = helpers.past_covs(
|
|
past_covariates, data, train_split, use_scalers
|
|
)
|
|
|
|
if past_covariates is not None:
|
|
lags_past_covariates = lags
|
|
else:
|
|
lags_past_covariates = None
|
|
|
|
reg_model = RegressionModel(
|
|
output_chunk_length=output_chunk_length,
|
|
lags=lags,
|
|
lags_past_covariates=lags_past_covariates,
|
|
)
|
|
|
|
with warnings.catch_warnings():
|
|
warnings.simplefilter(action="ignore", category=FutureWarning)
|
|
reg_model.fit(ticker_series, past_covariate_whole)
|
|
|
|
# Showing historical backtesting without retraining model (too slow)
|
|
return helpers.get_prediction(
|
|
"Regression",
|
|
probabilistic,
|
|
use_scalers,
|
|
scaler,
|
|
past_covariates,
|
|
reg_model,
|
|
ticker_series,
|
|
past_covariate_whole,
|
|
train_split,
|
|
forecast_horizon,
|
|
n_predict,
|
|
)
|