Hello everyone,

I’m currently working on a time-series forecasting project and have encountered an issue with missing data. I’m using historical sales data, and there are a few gaps in the time series—some periods have missing values due to system errors or incomplete data reporting. I’m wondering what the best approach is to handle these missing values without compromising the integrity of my model.

Specifically, I’m looking for advice on:

Imputation methods: Should I use forward filling, linear interpolation, or other imputation techniques? Which one works best for time-series data? Model impact: How do missing data points affect the performance of forecasting models like ARIMA, Prophet, or LSTM? Should I exclude the gaps or is there a better way to incorporate them? Best practices: Are there any general best practices for working with missing data in time-series analysis that I should be aware of? I’d really appreciate any guidance, particularly if you’ve encountered similar issues in your own projects. Looking forward to your thoughts!

Best, ragdoll hit