if ther was unexistent days/gaps on train&test data, do it have an impact on correct output when working time series forecasting.
yes correct i think - it will.
but should i create some scheme to fill that data or i could left this as it is (for example prices days: 1 2 3 5 ← no 4 day)
Missing or nonexistent data in time series forecasting can impact the accuracy of the output. It is recommended to address missing data before training the model.
You can consider techniques like Last Observation Carried Forward, Next Observation Carried Backward, Rolling Statistics (i.e. Simple Moving Average, Weighted Moving Average) and data interpolation to fill in the gaps and maintain the continuity of the time series.
Thank you!
1 Like