Machine-Learning

Machine Learning Projects

Topic 0 - Time Series EDA - Exploration / Transformation Techniques

with 5 univariate time series datasets - Chocolate Production, JetRail, Air Passengers, Wine Sales, and Portland Riders

1. Time Series EDA Univariate (Python)

2. Features extration with Tsfresh (Python)

BONUS 1) [Data Cleaning] - WIP

Topic 1 - Simple / Classical Time Series Forecasting Techniques

with 3 univariate time series datasets - JetRail (by day), Air Passengers (by month - multiplicative), and Wine Sales (by month - additive)

with 3 multivariate time series datasets for VAR models - TBD

0. [Summary - All Simple Models in One Notebook Comparison (Python)] - WIP

1. Naive Forecasting Approach (Python)

2. Simple Average Forecasting Approach (Python)

3. Moving Average (MA) Forecasting Approach (Python)

4. Simple Exponential Smoothing (SES) Forecasting Approach (Python)

5. Holt’s Linear Trend Forecasting Approach (Python)

6. Holt Winter’s Exponential Smoothing (HWES) Forecasting Approach (Python)

7. Autoregression (AR) Forecasting Approach (Python)

8. Autoregressive Moving Average (ARMA) (Python)

9. Autoregressive Integrated Moving Average (ARIMA) Approach (Python)

10. [Seasonal Autoregressive Integrated Moving-Average (SARIMA) Approach (Python)] - WIP

11. [Seasonal Autoregressive Integrated Moving-Average with Exogenous Regressors (SARIMAX) Approach (Python)] - WIP

12. [Vector Autoregression (VAR) Approach (Python)] - WIP

13. [Vector Autoregression Moving-Average (VARMA) Approach (Python)] - WIP

14. [Vector Autoregression Moving-Average with Exogenous Regressors (VARMAX) Approach (Python)] - WIP

15. [Forecast Combinations Approach (Python)] - WIP

16. Seasonal Naive Approach (R) - WIP

BONUS 1) [All Simple Models with R] - WIP

Topic 2 - Advanced Time Series Forecasting Techniques

1. Auto-Arima (R)

2. Auto-Arima (Python) with Pyramid

3. LSTM Time Series with Keras

4. LSTM Time Series with PyTorch

5. Boosting Algorithm - Adaboost Regressor (Python) with Tsfresh

6. Temporal HIErarchical Forecasting with R

Topic 3 - Applied Time Series Business Case - Fisheries and Environmental Data

Credit to E. E. Holmes, M. D. Scheuerell, and E. J. Ward - NWFSC: Northwest Fisheries Science Centern (https://nwfsc-timeseries.github.io/atsa-labs/)