nankezi1 / aolfortimeseries

Experiments for our paper "Adaptive Online Learning for Time Series Prediction".

Adaptive Online Time Series Prediction

How to run this code

Install Python 3.8.1 and Numpy version 1.18.1

• clone this project

  git clone https://github.com/OnlinePredictor/AdaptiveOnlineTimeSeriesPrediction.git

• to run the experiments later you need to install joblib

  pip install joblib

Overview

This repository contains different python files.

These files define our problems we used to test our algorithm.

1. RevisionExperimentSetting1.py
2. RevisionExperimentSetting2.py
3. RevisionExperimentSetting3.py
4. RevisionExperimentSettingReal1.py
5. RevisionExperimentSettingReal2.py
6. RevisionExperimentSettingReal3.py

The following files contain the predictors.

1. ArimaAutoregressor.py
2. Jamil_algo1.py
3. ARIMA_AO_Hedge.py

Optimizers:

1. RealOGD.py
2. RealONS.py
3. ADAFtrl.py
4. Ftrl4.py

Run experiments

To reproduce the results in our paper run RunExperiments.py. python PaperExperiments.py Running the experiments like this will start many threads and require a lot of computational power. If you do not want to run all experiments adjust the tasklist array in the main method.