The ultimate course to master time series forecasting in Python
By the end of this course, you will be able to tackle any forecasting problem with confidence and have a portfolio of 16 forecasting projects!
In this 13 hour course, you will:
- Define time series data
- Learn when forecasting not applicable (random walk model)
- Forecast with the SARIMAX family of models (model seasonal data and add exogenous variables)
- Forecast with exponential smoothing
- Perform multivariate forecasting
- Learn to deal with multiple seasonal periods (using BATS and TBATS)
- Forecast using decomposition (Theta model)
- Develop a framework to apply deep learning models like DNNs, LSTMs, and CNNs
- Work with the automated forecasting library Prophet
- Apply state-of-the-art forecasting models:
- N-BEATS
- N-HiTS
- PatchTST
- TimesNet
- TimeGPT
- and more to come in future updates
- Predict sparse time series
- Learn to use machine learning (XGBoost, LightGBM, etc.) to efficiently forecast time series
- Apply deep learning at scale and learn to work with large datasets, run efficient hyperparameter optimization and generate prediction intervals with deep learning models
Starter notebooks are available to follow easily along and all solutions are provided on GitHub!