Becoming a Good Data Science Customer

Course details

The world of data science has created radical shifts in the way we work, play, shop, socialize, and learn. And as it continues to grow in importance across many different industries, business leaders need to know how to communicate effectively with data science teams to maximize their return on investment. In this course, Columbia University professor Howard Friedman gives you an overview of the most critical questions and tools to probe data scientists about key topics related to data collection, data storage, data analysis, hardware and software, data modeling, ethics, and more. Along the way, explore the basic technical lingo, recognize the types of talent on the team, and learn how to pose well-formed questions to data scientists to gather better insights, create opportunities, and generate value as well as challenge key assumptions. By the end of this course, you’ll be equipped with the right questions and tools to start making more profitable data-driven business decisions.

Concepts

0. Introduction

01 - Good data science customers ask critical questions
02 - What you should know for this course

1. Tools of the Trade

03 - Stages of data workflow
04 - Data storage options
05 - Data sources
06 - Ensuring data quality
07 - Coding languages and repositories
08 - Data products
09 - Data workflow exercise

2. Descriptive Statistics Foundations

10 - Data limitations
11 - Summary statistics
12 - Correlations
13 - Testing hypothesis - Effect sizes and p-values
14 - Common mistakes

3. Making Good Decisions with Data

15 - Causality
16 - Benefits of randomization
17 - What if you can't randomize
18 - Spotting biases
19 - Matching description to bias exercise

4. Customer Segmentation

20 - What projects involve unsupervised machine learning
21 - Reducing dimensions
22 - Clustering algorithms

5. Predictive Modeling

23 - What projects involve predictive modeling
24 - Feature selection
25 - Model training and testing (data partitioning)
26 - Model tuning
27 - Measuring model performance
28 - Confusion matrix
29 - Customer simulation exercise

6. Other Modeling Methods

30 - Natural language processing (NLP)
31 - Geospatial analysis
32 - Computer vision
33 - Network analysis
34 - Large language models (LLM)

7. Ethics

35 - Data science code of ethics
36 - Bias and fairness
37 - Data drift
38 - Privacy and security

Conclusion

39 - Next steps