Course curriculum

    1. Learning Objectives

    2. About the Instructor

    3. Predictions vs Decisions

    4. Intro to Profit Curve

    5. The Data And The Model

    6. Profit Curve

    7. Finding the Optimal Threshold

    8. Profit Curves for Prioritization

    9. Model Improvements

    10. Profit Curves Biggest Limitation

    11. Flexible Thresholds

    12. Chapter Quiz

    1. Introduction and the Grupo Bimbo Problem

    2. Answer to the Inventory Management Thought Experiment

    3. How Mean Absolute Error Works

    4. Comparing MAE and MSE in Code

    5. Model and Decision Infra Code

    6. Model and Stocking Metrics

    7. Comparing Stocking Rules

    8. Alternative Loss Function

    9. More Direct Optimization

    10. Optimization With Full Distributions

    11. Chapter 2 Quiz

    1. Static vs Dynamic Decision Making - An Example

    2. Static vs Dynamic Decision Making - Conceptual Framework

    3. Choosing Between the Static vs Dynamic Framework

    4. Data and Problem Prep for Dynamic Optimization

    5. Connecting the Predictive Model to Simulation loop

    6. Validating business rules

    7. The simulation loop

    8. Running the Decision Rule Comparison

    9. Viewing Comparison Results

    10. Sweeps for programmatic optimization

    11. Simulation with multiple ML models

    12. Chapter 3 Quiz

    1. Understanding Drift: Concept Drift vs. Covariate Shift

    2. Challenges in Decision Optimization: Drift, Sim-to-Real, and Testing Decisions

    3. Detecting and Addressing Drift in Data: Techniques and Tools

    4. Reporting Drift Patterns

    5. Understanding Covariate Shift and its Implications for Predictive Models

    6. Simple Concept Drift Adjustments

    7. Adjusting Concept Drift with SHAP Values

    8. The Sim2Real Problem

    9. Testing Randomization vs Contamination

    10. The Simulate - Optimize - Test Cycle

    11. Closing Thoughts and Further Resources

    12. Course Assignment

About this course

  • Free
  • 47 lessons
  • 4 hours of video content

Learn to optimize decision rules, translating machine learning predictions into actionable insights. Discover how to achieve practical value and business impact by measuring performance using business metrics, and deploy ML models successfully.

Sign up for this free Weights & Biases course to:

  • Translate Machine Learning Predictions Into Actionable Business Insights

    Learn how to identify decision and prediction problems in your work and optimize them using advanced machine learning models. Understand the significance of transforming abstract data into direct, actionable insights that can power your business decisions and strategies.

  • Master Business Impact Measurement for ML Models

    Instead of relying solely on accuracy metrics, you'll learn how to measure the performance of your ML models in dollar terms. This will enhance your ability to drive impactful results and high return on investment from your ML initiatives.

  • Choose the Right Loss Function for Your Business Problem

    Enhance your proficiency in selecting the most appropriate loss function for your specific business problem. This skill will empower you to implement ML models more effectively, ensuring they are attuned to your business objectives, thereby increasing their relevance and impact.

Instructor

Dan Becker

Dan Becker started working as a data scientist after finishing in 2nd place (out of 1350 teams) in a Kaggle competition with a $500,000 grand prize. Since then, he has led consulting projects for 6 companies in the Fortune 100, contributed to open source projects like TensorFlow and worked as a data scientist at Google. Dan founded Decision AI in 2020 to help data scientists optimize how they turn predictions from machine learning models into better business decisions. Dan sold Decision AI to DataRobot where he led the product team building ML Development tools. Dan has a PhD in Economics and he remains focused on combining ML and economic optimization techniques to make machine learning as practically valuable as possible.

Prerequisites

  • Basic knowledge of machine learning

  • Familiarity with Python programming

Course Reviews

Hear from Business Decision Optimization Certification takers

5 star rating

It is insighful to connect ML to something bit more tangible

Vaclav Kosar

Enroll for free now and earn your certificate