Time series forecasting

Time series forecasting

• LSTM (Long Short Term Memory)

• ETS (Exponential smoothing state space)

• STL (Seasonal and Trend decomposition using Loess)

• Prophet

• Decision Tree Regressor

Configuration

Input data: 
ts_df            # dataframe['timestamp', 'kpivalue']
ts_sample_frequency = '60min'  # resample time window

Training data:
ts_df_scaler                  # scaler of kpivalue
ts_features = ['hourly', 'weekly', ...]  # using time feature for training
train_size_rate = 0.7       # split training and testing data

Project

set 'lstm-time-series' as working directory

• In Pycharm

in pycharm > file > setting > console > working directory )

• In python script

` import os

os.chdir("C:\...\lstm-time-series")

print(os.getcwd()) `