Using LSTM RNN to forecast time series; includes sine wave, electrocardiogram and ad impression forecasting
Slideshow presentation:
https://docs.google.com/presentation/d/16BKVKNZbm0HJUiFl8UeZDi9HjD4OmqfvoKmTMMK78Nk/edit?usp=sharing
FILES
model.py: setup for training (file will work for all 3 experiments)
Sine Wave Forecasting
sine.py: TensorFlow Training Model, plotting
ECG Forecasting
ecg.py, TensorFlow Training Model, plotting
ECGdata.py (computer generated ECG data)
Ad Impression Forecasting
ads.py: TensorFlow Training Model, plotting
testAdID1.py, testAdID3.py, testAdID3.py: test data sets of ad impressions
TRY
Adjusting Hyperparameters:
-increase & decrease RNN LAYERS, TIMESTEPS, LEARNING RATE, etc