Using Google Search Trends to Estimate Global Patterns in Learning
summary
Uses Google Trends plus a minimum mean square error estimator to reverse engineer per-country computer science literacy, producing a CSLI-s score scaled 0–1 by multiplying estimated search-depth with internet access rates. Validates against GitHub repo counts, OECD PISA scores and a global survey and finds positive correlations, year‑over‑year increases and curriculum clusters. Detects a 1.7% drop of a year’s CS search volume in the three weeks after COVID lockdowns. Notes key limits: access assumptions, cultural query habits and non‑Latin script challenges.
AI-written summary · about
chapters (13)
- 0:00Introduction and motivation
- 1:05Limitations of traditional learning metrics
- 2:06Google Trends as a learning signal
- 3:08Theoretical foundation for CS literacy
- 4:30Methodology overview
- 5:53Processing Google Trends data
- 7:38Topic selection and validation
- 8:41Results and correlation findings
- 9:23Regional trends and clustering
- 10:27Limitations and future work
- 11:07Impact of COVID-19 on learning
- 12:11Seasonal and curriculum patterns
- 13:13Caution on biases and next steps
related talks
44mAutomating Analysis
18mHow Data Science Can Unlock Teaching & Learning at Scale
40mData Doesn't Lie: How to Influence Open Source Programming Through..
01h 40mAfternoon Session #1: Science, Pseudo-science, Canon and Belief
47mExploring StackOverflow Data
37mAI/ML Strategies & Techniques for a Post Pandemic World
58mA future of data science
35mExposing Algorithms