Sridhar Telidevara: Professor, Great Lakes Institute of Management at Cypher 2017 - Part 1/2
summary
Shows how parametric and machine‑learning models produce interpretable predictions for crime and recidivism, using spatial‑temporal features and simple inputs to inform sentencing, probation and police beats. Demonstrates heat maps from retail geolocation (Amazon purchase density) as proxy indicators, models crime clustering with self‑exciting point process (aftershock) models, computes spatial autocorrelation via weight matrices, and uses standard GIS files (SHP/DBF/SHX) and tools like GeoDa/CrimeStat to generate hotspots and short‑term tactical forecasts. Results claim actionable hotspots and recidivism scores ready for operational use.
AI-written summary · about
chapters (11)
- 0:00Start
- 7:12What Is the Reason for Committing Crime
- 9:35The Rational Choice Theory
- 10:36Social Disorganization Theory
- 12:31Why Is It Important
- 14:18Spatial Temporal Analysis
- 15:29Takeaways
- 16:10Hot Spots of Crime
- 19:23Earthquakes in San Francisco
- 21:48Spatial Temporal Modeling of Time
- 22:57What Is a Hot Spot
related talks
47mKeynote: Using Data Science for Social Good: Examples, Opportunities, and Challenges
55mEthical Machine Learning
24mKeynote: Digital Echoes: Understanding Mass Violence with Data and Statistics
30mLocation Intelligence and Spatial Data Science: from geographic coordinates to business insights
47mAI Ethics, Strategic Decisioning and Game Theory
30mUnlocking Geospatial Data Science: A Strategic Overview
24mHunting Criminals with Hybrid Analytics
24mBuilding Aggregated Solutions to Scale Your Geospatial Strategies