note that this is just an exercise

i’m trying to use time series to predict btc of the next 100 days

pycaret 3.3.2 using jupyter notebook

import httpx 
import pandas as pd

# we get value of BTC during the last 500 days
req = httpx.get('https://min-api.cryptocompare.com/data/histoday?fsym=BTC&tsym=EUR&limit=500') 
# we convert that data from json to dataframe
rq_data = req.json()['Data'] 
data = pd.DataFrame(rq_data) 
# we clean the data so we use only what it appears interesting 
data = data[['time', 'high', 'low', 'open', 'volumefrom', 'volumeto', 'close']] 

"""we take 400 values from data , we train on 350 and test our model on 50... so later we can predict 100 next days and compare the prediction with the real value (500) to check if our model is good"""
from pycaret.time_series import * 
s = setup(data.head(400), fh = 50, target='close', session_id = 123) 

best = compare_models() 
plot_model(best, plot = 'forecast') 

# everything works except this last command, the goal is to predict the value of BTC during the next 100 days 
plot_model(best, plot = 'forecast', data_kwargs = {'fh' : 100}) 

# it gives this error: 

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[10], line 2
 1 # plot forecast for 36 months in future
----> 2 plot_model(best,  plot  =  'forecast',  data_kwargs  =  {'fh'  :  100})

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/utils/generic.py:964, in check_if_global_is_not_none.<locals>.decorator.<locals>.wrapper(*args, **kwargs)
 962     if globals_d[name] is None:
 963         raise ValueError(message)
--> 964 return func(*args,  **kwargs)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/time_series/forecasting/functional.py:1400, in plot_model(estimator, plot, return_fig, return_data, verbose, display_format, data_kwargs, fig_kwargs, save)
 1163 @check_if_global_is_not_none(globals(), _CURRENT_EXPERIMENT_DECORATOR_DICT)
 1164 def plot_model(
 1165     estimator: Optional[Any] = None,
 (...)
 1173     save: Union[str, bool] = False,
 1174 ) -> Optional[Tuple[str, list]]:
 1175  """
 1176     This function analyzes the performance of a trained model on holdout set.
 1177     When used without any estimator, this function generates plots on the
 (...)
 1397 
 1398     """
-> 1400     return _CURRENT_EXPERIMENT.plot_model(
 1401  estimator=estimator,
 1402  plot=plot,
 1403  return_fig=return_fig,
 1404  return_data=return_data,
 1405  display_format=display_format,
 1406  data_kwargs=data_kwargs,
 1407  fig_kwargs=fig_kwargs,
 1408  save=save,
 1409  )

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/time_series/forecasting/oop.py:4198, in TSForecastingExperiment.plot_model(self, estimator, plot, return_fig, return_data, verbose, display_format, data_kwargs, fig_kwargs, save)
 4195 system = os.environ.get("PYCARET_TESTING", "0")
 4196 system = system == "0"
-> 4198 return self._plot_model(
 4199  estimator=estimator,
 4200  plot=plot,
 4201  return_fig=return_fig,
 4202  return_data=return_data,
 4203  verbose=verbose,
 4204  display_format=display_format,
 4205  data_kwargs=data_kwargs,
 4206  fig_kwargs=fig_kwargs,
 4207  save=save,
 4208  system=system,
 4209 )

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/time_series/forecasting/oop.py:3758, in TSForecastingExperiment._plot_model(self, estimator, plot, return_fig, return_data, verbose, display_format, data_kwargs, fig_kwargs, system, save)
 3753 X = data_kwargs.get("X", None)
 3754 return_pred_ints = [
 3755     estimator.get_tag("capability:pred_int") for estimator in estimators
 3756 ]
-> 3758 model_results = [
 3759     self.predict_model(
 3760         estimator,
 3761         fh=fh,
 3762         X=X,
 3763         return_pred_int=return_pred_int,
 3764         alpha=alpha,
 3765         coverage=coverage,
 3766         verbose=False,
 3767     )
 3768     for estimator, return_pred_int in zip(estimators, return_pred_ints)
 3769 ]
 3771 if len(estimators) == 1:
 3772     return_pred_int = return_pred_ints[0]

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/time_series/forecasting/oop.py:3759, in <listcomp>(.0)
 3753 X = data_kwargs.get("X", None)
 3754 return_pred_ints = [
 3755     estimator.get_tag("capability:pred_int") for estimator in estimators
 3756 ]
 3758 model_results = [
-> 3759     self.predict_model(
 3760  estimator,
 3761  fh=fh,
 3762  X=X,
 3763  return_pred_int=return_pred_int,
 3764  alpha=alpha,
 3765  coverage=coverage,
 3766  verbose=False,
 3767  )
 3768     for estimator, return_pred_int in zip(estimators, return_pred_ints)
 3769 ]
 3771 if len(estimators) == 1:
 3772     return_pred_int = return_pred_ints[0]

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/time_series/forecasting/oop.py:4567, in TSForecastingExperiment.predict_model(self, estimator, fh, X, return_pred_int, alpha, coverage, round, verbose)
 4564 fh = self._predict_model_reconcile_fh(estimator=estimator_, fh=fh)
 4565 X = self._predict_model_reconcile_X(estimator=estimator_, X=X)
-> 4567 result = get_predictions_with_intervals(
 4568  forecaster=pipeline_with_model,
 4569  alpha=alpha,
 4570  coverage=coverage,
 4571  X=X,
 4572  fh=fh,
 4573  merge=True,
 4574  round=round,
 4575 )
 4576 y_pred = pd.DataFrame(result["y_pred"])
 4578 #################
 4579 # Metrics ####
 4580 #################

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pycaret/utils/time_series/forecasting/__init__.py:110, in get_predictions_with_intervals(forecaster, alpha, coverage, X, fh, merge, round)
 108 # Get Point predictions ----
 109 if alpha is None:
--> 110     y_pred = forecaster.predict(fh=fh,  X=X)
 111     y_pred = pd.DataFrame({"y_pred": y_pred})
 112 else:

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/base/_base.py:431, in BaseForecaster.predict(self, fh, X)
 429 # we call the ordinary _predict if no looping/vectorization needed
 430 if not self._is_vectorized:
--> 431     y_pred = self._predict(fh=fh,  X=X_inner)
 432 else:
 433     # otherwise we call the vectorized version of predict
 434     y_pred = self._vectorize("predict", X=X_inner, fh=fh)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/compose/_pipeline.py:521, in ForecastingPipeline._predict(self, fh, X)
 506 """Forecast time series at future horizon.
 507 
 508 Parameters
 (...)
 518     Point predictions
 519 """
 520 X = self._transform(X=X, y=fh)
--> 521 return self.forecaster_.predict(fh,  X)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/base/_base.py:431, in BaseForecaster.predict(self, fh, X)
 429 # we call the ordinary _predict if no looping/vectorization needed
 430 if not self._is_vectorized:
--> 431     y_pred = self._predict(fh=fh,  X=X_inner)
 432 else:
 433     # otherwise we call the vectorized version of predict
 434     y_pred = self._vectorize("predict", X=X_inner, fh=fh)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/compose/_pipeline.py:1032, in TransformedTargetForecaster._predict(self, fh, X)
 1017 def _predict(self, fh=None, X=None):
 1018  """Forecast time series at future horizon.
 1019 
 1020     Parameters
 (...)
 1030         Point predictions
 1031     """
-> 1032     y_pred = self.forecaster_.predict(fh=fh,  X=X)
 1033     # inverse transform y_pred
 1034     y_pred = self._get_inverse_transform(self.transformers_pre_, y_pred, X)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/base/_base.py:431, in BaseForecaster.predict(self, fh, X)
 429 # we call the ordinary _predict if no looping/vectorization needed
 430 if not self._is_vectorized:
--> 431     y_pred = self._predict(fh=fh,  X=X_inner)
 432 else:
 433     # otherwise we call the vectorized version of predict
 434     y_pred = self._vectorize("predict", X=X_inner, fh=fh)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/base/adapters/_pmdarima.py:108, in _PmdArimaAdapter._predict(self, fh, X)
 106 # all values are out-of-sample
 107 if fh.is_all_out_of_sample(self.cutoff):
--> 108     y_pred = self._predict_fixed_cutoff(fh_oos,  X=X)
 110 # all values are in-sample
 111 elif fh.is_all_in_sample(self.cutoff):

File ~/Desktop/Python/venv/lib/python3.10/site-packages/sktime/forecasting/base/adapters/_pmdarima.py:218, in _PmdArimaAdapter._predict_fixed_cutoff(self, fh, X, return_pred_int, alpha)
 203 """Make predictions out of sample.
 204 
 205 Parameters
 (...)
 215 Returns series of predicted values.
 216 """
 217 n_periods = int(fh.to_relative(self.cutoff)[-1])
--> 218 result = self._forecaster.predict(
 219  n_periods=n_periods,
 220  X=X,
 221  return_conf_int=False,
 222  alpha=DEFAULT_ALPHA,
 223 )
 225 fh_abs = fh.to_absolute(self.cutoff)
 226 fh_idx = fh.to_indexer(self.cutoff)

File ~/Desktop/Python/venv/lib/python3.10/site-packages/pmdarima/arima/arima.py:787, in ARIMA.predict(self, n_periods, X, return_conf_int, alpha, **kwargs)
 785 X = self._check_exog(X)  # type: np.ndarray
 786 if X is not None and X.shape[0] != n_periods:
--> 787     raise ValueError(
 788         f'X array dims (n_rows) != n_periods. Received '
 789         f'n_rows={X.shape[0]} and n_periods={n_periods}'
 790     )
 792 # f = self.arima_res_.forecast(steps=n_periods, exog=X)
 793 arima = self.arima_res_

ValueError: X array dims (n_rows) != n_periods. Received n_rows=50 and n_periods=100

i tried to follow the official tutorial , making some adjustments for my exercise

i expected that it predict the value of btc for next 100 days, means from day 400 to 500 (so i can compare it with the original plot)