Open Mohan16071996 opened 1 month ago
can anyone please suggest on how to fix this issue?
Also, my data set has only business days and also holidays and weekend data are not available. Maybe the missing combination is because of that?
Which of the suggestions from the error message did you try?
What happened + What you expected to happen
1.the bug: We are using Informer model with neural forecast. I get the mentioned error when I predict. forecasts = nf.predict(futr_df=test_data)
futr_df
.\n" 753 f"You can run the{expected_cmd}
method to get the expected combinations or "ValueError: There are missing combinations of ids and times in
futr_df
. You can run themake_future_dataframe()
method to get the expected combinations or theget_missing_future(futr_df)
method to get the missing combinations.useful info: test_data size is 77*13, training data size is 26246 rows × 13 columns, which contains additional features added. I followed the sample informer given. input size is given as 77, the op is 7 days, for 11 tickers.
Versions / Dependencies
In Google Colab.
Reproduction script
Config:
model = Informer( h=7, # Predicting for 7 days input_size=77, # Use 14 days of history hidden_size=16, conv_hidden_size=32, n_head=2, loss=MAE(), futr_exog_list=calendar_cols, scaler_type='robust', learning_rate=1e-3, max_steps=1000, val_check_steps=50, early_stop_patience_steps=10 )
Pre processing: combined_data_panel, calendar_cols = augment_calendar_df(df=combined_data, freq='B')
Split data into train and test sets
train_data = combined_data_panel[combined_data_panel.ds < combined_data_panel['ds'].values[-7]] test_data = combined_data_panel[combined_data_panel.ds >= combined_data_panel['ds'].values[-7]].reset_index(drop=True)
train_data['ds'] = pd.to_datetime(train_data['ds']) test_data['ds'] = pd.to_datetime(test_data['ds'])
nf.fit(df=train_data, static_df=df_one_hot, val_size=12)
forecasts = nf.predict(futr_df=test_data)
Issue Severity
High: It blocks me from completing my task.