I wanted to see a notebook / tutorial that would take me through the basics of working with time series.
Most notebooks I saw were either not very rigorous or they took me straight into price prediction using some methods, which is not what I needed right away.
So I wrote my own introductory notebook. I acquired most of the information I used here through reading Introduction to Time Series Forecasting With Python by Jason Brownlee.
What this notebook is:
- a good starting point for understanding time series data and how it differs from problems with other type of tabular data
- a cookbook to use for exploration when starting to work with a new dataset
What this notebook is not:
- it is not about prediction. It stops at exploration and understanding the data.
- it's not meant for advanced practitioners of asset price prediction - unless you want to revisit some concepts.
Contents
- Quick overview
- Dataset description
- Basic trading data visualization
- Preprocessing
- Feature engineering
- Typical time series visualizations
- Power transforms
- Temporal structure of time series data
- Model evaluation