# How to model a time series problem for RNNs?

I have projects as input data, each project has a weekly progress report (hours of work completed in this week). A project can have an arbitrary duration, but let's say it's usually around 100 weeks, so 100 timesteps for each project. For every project, there is a fix amount of hours that we know from the beginning. Whenever this amount of total hours is reached, the project is considered completed.

Assuming I am in the middle of a project, I would like to use all data I already have to forecast how many weeks it takes for the project to complete.

Since we are talking about a time series data, I thought it might be a good idea to use a RNN (specifically LSTM). But if I use this, what would the input data look like? Should I just the features prev_week_hours and the labels current_week_hours? Or I could also use more timesteps, like prev_week_hours, prev_prev_week_hours, .... However, is this even necessary for RNNs? or do they automatically caputure these timesteps?

In conclusion, I am really not sure about this: How can I can properly model this problem to be able to use machine learning algorithms to forecast project completion?