[BUG] ARIMA future covariates not working

[Japsz]

Describe the bug
ARIMA model is not letting me introduce exogenous variables into the model.

To Reproduce

df = pd.DataFrame([[1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [2,3], [3,4], [4,5], [5, 4], [3, 5], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4], [4,5], [5, 4], [3, 5], [1,2], [2,3], [3,4`], [4,`5], [5, 4], [3, 5]], columns=['target', 'covar'])
df.index = pd.RangeIndex(start=0, stop=len(df))

series = TimeSeries.from_dataframe(df, None, ['target', 'covar'])
train, test = train_test_split(series, test_size=0.2)

arima = ARIMA(p=1, d=1, q=1)
arima.fit(train['target'], future_covariates=series['covar'])
preds = arima.predict(len(test['target']), future_covariates=series['covar'])

[2022-03-11 17:07:28,770] ERROR | darts.models.forecasting.forecasting_model | ValueError: For the given forecasting horizon 'n=10', the provided 'future_covariates' series must contain at least the next 'n=10' time steps/indices after the end of the target 'series' that was used to train the model.

Expected behavior
I'm expecting this to work since I'm making sure that there's already at least n point after the training series ends.

System (please complete the following information):

• Python version: 3.8.5
• darts version 0.17.1

Additional context
I've been debugging a lot and seems that the internal slicing will always give a future covariate series that's one less than the prediction needs (when freq is 1), hence throwing the error.
in darts/models/forecasting/forecasting_model.py line 1183

future_covariates = future_covariates[
                start : start + (n - 1) * self.training_series.freq
            ]`

[Japsz]

Can someone confirm if this is intended or not? Just wanna make sure I'm not doing something wrong with my code

[dennisbader]

Hi @Japsz and thanks for pointing this out.

You are right, the slicing only correctly works for DatetimeIndex and not a RangeIndex.

We need to fix this

[DeastinY]

I've run into the same error today. Good timing 😁

[parshinsh]

I'm still getting the error of RangeIndex for ARIMA and VARIMA.

[dennisbader]

Which darts version are you using?
Please share the error message along with a minimal code snippet to reproduce the issue, thanks!

[yigitcancomlek]

Seems like this issue still persists for the ARIMA (ARMA) Model. I am still getting the same error with the below code

train_ts, val_ts= target_ts.split_before(train_cutoff) #40 samples for training # 150 samples for validation
train_val_cov  = covariates_ts; # all training and validation covariates (190x4)

model = ARIMA(p=1,d = 0,q=1)

model.fit(series = train_ts,
              future_covariates = covariates_ts)

predictions = model_ts.predict(n=150
              series=train_ts,  
              future_covariates=covariates_ts,
              num_samples = 100)

gives me the below error. Please let me know if I am making any mistake but if not, I believe the issue still persists. I have u8darts-all 0.24.0 downloaded in my conda environment

ValueError: Provided exogenous values are not of the appropriate shape. Required (110, 9), got (150, 9).