Moving Average - Heisenberg0203/GETS GitHub Wiki

Moving Average (MA) takes the mean of several historical observations to predict future values. The forecaster uses data from the current period to previous N observation depending on the window width. The varying pattern in the time series data affects the width of the window and the amount of smoothing required to make predictions.

Forecast equation for MA:


forecast_t+k = (W_t-1 + W_t-2 + W_t-3 + .... W_t-n) /N

Grammar for MA:

The grammar for Moving Average is shown in where <window_var> generates the width of the window and <lag_var> generates the lag in time series which is implemented using the shift function provided in Python’s Pandas library to forecast by substituting parameters in the above equation.

Sample Derivation Tree:

Executing MA ( Please ensure that you are in src folder)

python ponyge.py --parameters gets/sma.txt

By default it runs on Daily Waste Generated dataset. Command to run on the custom dataset:

python ponyge.py --parameters gets/sma.txt --dataset_train path_to_traindataset --dataset_test path_to_testdataset.