The 80/20 Rule of Data Science

1h 19mIntermediate2024-09-04

Authors

Howard Friedman

Course details

In this course, join instructor Howard Friedman as he outlines the core concepts of the 80/20 rule for data science and how to get most of the value with minimal effort. Explore the basic steps required to assign value to different parts of your work, including defining success, obtaining high-quality datasets, starting with basic statistics, building basic models early, and avoiding getting bogged down by less impactful details. This course is targeted at data team leaders and senior data team members in the tech world who are interested in being efficient and effective business partners.

Skills covered

Data Science FoundationsManagement SkillsData AnalysisData ScienceBusiness Analysis and StrategyLeadership and ManagementBusiness Software and ToolsOne-Off

Concepts

0. Introduction

01 - What is the 80 20 rule of data science
02 - What you should know for this course

1. Getting Started with Prioritizing

03 - Defining success
04 - Make minimum viable products (MVP) your friend
05 - The point of diminishing returns
06 - Why do data science projects fail
07 - Plan for deployment

2. Database Selection and Quality Checks

08 - Comb the literature
09 - Triaging data sources
10 - Checking data quality
11 - Dealing with missing data
12 - Finding strong features

3. Simple Models First

13 - Feature transformations
14 - Transforming continuous features
15 - Interaction terms
16 - Adding data sources
17 - Linear regression
18 - Logistic regression

4. Tracking Performance

19 - Choosing metrics
20 - Designing simple dashboards
21 - Rapid prototyping
22 - When to stop

5. Communicating the 80 20 of Data Science

23 - Focus on outcome, not technique
24 - Manage expectations and build trust
25 - Emphasize collaboration

Conclusion

26 - Key takeaways for your future career in data science