Time series regression question

Question

I'm wondering if someone could help. I've not really created a time series regression model before and would like to be pointed in the right direction if possible.

I have a number of continuous independent variables which I have recorded over 6 time points (months), and a binary dependent variable which indicates whether an event occurred or not, plus the time point that this event occurred. I have a set of 50,000 records, 9% of these have an event flagged at some point in the time series.

My brief aim is to create a model which could give me a probability of an event occurring next month, based on this time series data, or something to this effect.

I will be using R to create my model so any packages or process recommendations would be ideal.

As I said, I'm a beginner with time series so please forgive my naivety, but if someone could give me a starting point or point me in the direction for more information that would be great and very much appreciated.

Answer 1

One approach that would work for you is to use a recurrent neural network.

The caveat is that those are easier to work with in Python than in R (There are R libraries for RNN, but personally I've had difficulties with them).

I'm pretty sure State Space Models and Dynamic Linear Models can also solve your problem, and for those there is ample R documentation available, but I don't have any direct experience with those approaches.