LSTM time-series forecasting over aggregated window features where each row represented as separated object

Question

I have dataset with window features of data usage.

[ W1, W2, W3, ... Wn ] [W0]

Where each W1..Wn - is aggregated sum of data usage per Month/Week/ W0 - is target data usage that model must predict.

The thing is that each row is a separated object - that represents only one particular client and there are no timestamp between the rows, only aggregated window information as column for different client.

How to use LSTM model for regression in such cases? Or is it better to use another approach?

Thank you.