This is my third model option for forecasting search interest for ten forms of outdoor recreation in Michigan. This approach uses a Long Short Term Memory (LSTM) model. This is a recurrent neural network that considers sequences of values while also considering a state when creating model predictions.

This model shows a lot of promise. In future installments, I will compare cross validation performance between this and my first model. My second model will be dropped from consideration due to similarity in results to the first model.

I will be focused on other things this month. However, I hope to wrap up and summarize this project tentatively sometime this fall.