Different sequences lenght in time series forecasting

Hello everyone!

I’m trying to run a model which predicts the river flow for next month (t+1) using a precipitation forecast model (also t+1) and observed data (t, t-1, …, t-n). So my dataset will be something like:

features: ‘precipitation forecast data (t+1), observed precipitation (t, … t-n), observed inflow (t, … t-n)’

output : ‘inflow forecast (t+1)’

Does anyone have a suggestion on how I shall build my DataLoader for the NN input data? I’m currently building a manual data set with one column for each data and time step. Kinda looking like this:

P (t+1) | P (t) | P (t-1) | P (t-n) | Inf (t) | Inf (t-1) | Inf (t-n)| Inf (t+1)

Thanks!