This course introduces investment professionals to the biases in data science pipelines and ways to mitigate them. You will see the big picture of ethical dilemmas, biases and practical issues encountered in finance-related data science projects. You will also dive into different stages in a data science project and take a closer look at the specific biases in each stage and discuss mitigation strategies. You will learn how practitioners take apart a machine learning model, identify potential biases, and take necessary measures to mitigate them.
This course is designed for those who want to understand how biases are identified and mitigated in finance-related data science projects, and is especially suited for the translator role, which is those individuals working with and communicating between both investment and data science teams.
Course modules include:
- Module 1: Investment Context, Some Ethical Dilemmas, Biases and Practical Issues
- Module 2: Biases and Mitigations in Study Design and Hypothesis Generation
- Module 3: Biases and Mitigations in Data Collection and Exploration
- Module 4: Biases and Mitigations for Model Development, Testing, and Monitoring
- Module 5: Biases and Mitigations in Model Interpretation, Communication, and Governance
- Module 6: Case Studies and Code Labs: Mitigating Biases in the Data Science Pipeline
Prerequisites
This course is part of a data science series. It is suggested, but not required, that learners complete the courses in the recommended order below to ensure uniform foundational knowledge.
Recommended order:
- Data and Statistics Foundation for Investment Professionals
- Statistics for Machine Learning for Investment Professionals
- Machine Learning for Investment Professionals
- Natural Language Processing for Investment Professionals
- Mitigating Biases in the Data Science Pipeline for Investment Professionals
