Nixtla
Query on forecast
fcst_df = nixtla_client.forecast(train, h=7, freq = 'B' , time_col = 'ds', target_col='Close', add_history=False,level=[80, 90])
If my training data has late date as of date : 2024-04-18
as I have given the add_history as false, I will get the fcst_df as this:
ds | TimeGPT | TimeGPT-lo-90 | TimeGPT-lo-80 | TimeGPT-hi-80 | TimeGPT-hi-90
2024-04-19 | 168.687851 | 165.588795 | 166.693799 | 170.681903 | 171.786906 2024-04-22 | 169.127777 | 162.418342 | 163.446799 | 174.808755 | 175.837212 2024-04-23 | 169.353394 | 160.870766 | 162.072232 | 176.634555 | 177.836021 2024-04-24 | 169.278168 | 159.715937 | 163.565688 | 174.990648 | 178.840398 2024-04-25 | 169.264404 | 162.180334 | 163.522729 | 175.006079 | 176.348475 2024-04-26 | 169.216583 | 161.188441 | 162.522191 | 175.910976 | 177.244725 2024-04-29 | 169.127563 | 159.865936 | 160.671568 | 177.583559 | 178.389191
Now, this is the predicted value for those dates? Or forecasted values for 7 business days later from what is in the ds column in the fcst_df? I mean, value of 2024-04-19 is 168.687851 or predicted value for 7 business days later from 2024-04-19 is 168.687851?
Because add_history is set to false the model will forecast the future values based on the training data up to the latest date available in your training dataset. So the value of 2024-04-19 is where the forecast starts from and is 168.686851.
@Mohan16071996 Apologies for the late reply. Confirming @khizer-kt answer - does that help you?
