Explainable Automated Machine-Learning for Forecasting Financial Asset Prices
In this introductory tutorial, we demonstrate basic techniques for building prediction models for financial asset prices. We start with sourcing online data for building a baseline time-series model, and then proceed to improve the model performance by iterative approaches that introduce feature engineering, hyperparameter search, record linkage, and interpretable machine-learning methods such as AutoML with LIME in Driverless.ai. We will evaluate model performance and if time permits, discuss simple investment techniques.
Length of Tutorial: 1 Hour
Materials Required: Laptop if participants would like to be hands-on.