How Best To Perform Group Time Series Split for Irregular Time Series that has Unequal Daily Observations?

[nemohoe]

Hi All --

I am really new to coding and can do basic python programming, however, this problem I would like some help with.

I am using XGBoost on a irregular time series. Observations are recorded usually every two weeks, from Spring to Fall for 10 years. On days where observations were recorded, there are typically 5 observations of my target variable's value. However, there are days where there are 10 recorded observations and other days where there are 4 observations recorded. Does anyone have any suggestions/code ideas on how to perform GroupTimeSeriesSplit on my particular data set? I could just drop observations from days where the total exceeds 4, however, dropping data in this instance would be a last ditch effort.

Any help is greatly appreciated!