SafeDining2

Recommender for safe restaurants

Transcript

1. Ahmad Haider Insight Data Science 2016
2. Personal Story What if I could find which restaurants are safe and which are not? Mugged @Parking lot ! L Hungry for Pizza! Closest pizza location
3. Makes Real-Time Recommendations for Safety Ratings Restaurants in a Neighborhood SafeDining
4. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location
5. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location
6. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location
7. I want to eat Pizza at 3+ rated restaurant within 2 miles of my location p1 p2 p3 p4 p5 p1 p2 p3 p4 p1 p2 p1 p2 p3 p4 p5 p6 p7 p8 p1 Crime Location Crime Probability
8. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location p1 p2 p3 p4 p5 p1 p2 p3 p4 p1 p2 p1 p2 p3 p4 p5 p6 p7 p8 p1 Crime Location Crime Probability 1. pRest 1. pRest 1. pRest 1. pRest pRest Crime Probability at restaurant
9. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location p1 p2 p3 p4 p5 p1 p2 p3 p4 p1 p2 p1 p2 p3 p4 p5 p6 p7 p8 p1 Crime Location Crime Probability 1. pRest 1. pRest 1. pRest 1. pRest pRest Crime Probability at restaurant pRest pRest pRest pRest pRest pRest pRest
10. I want to eat Pizza at 2+ rated restaurant within 3 miles of my location p1 p2 p3 p4 p5 p1 p2 p3 p4 p1 p2 p1 p2 p3 p4 p5 p6 p7 p8 p1 Crime Location Crime Probability 1. pRest 1. pRest 1. pRest 1. pRest pRest Crime Probability at restaurant pRest pRest pRest pRest pRest pRest pRest 2. Relative Safety Index 2. Relative Safety Index 2. Relative Safety Index 2. Relative Safety Index
11. Workflow Crime dataset Yelp dataset Preprocessing Defining classification problem Feature Engineering Choosing a model Validation Python Pandas Regular expr. Multiclass 24 classes (hour of day) Standardization PCA Logistic Regression scikit-learn 10-fold cross validation Log loss score 2009-2015 Yelp search API
12. ™  Multiclass Classification –  Predict the hour at which crime happens at given location –  Features: Location (lat., long.), Address, Day, Week, Month, Year –  Labels: 24 classes (hour of day) –  Logistic Regression –  Cross entropy loss measure = 2.95 Algorithm
13. Theft Residential Burglary Robbery Assault Source: mylocalcrime.com Validation My analysis Independent source
14. •  Ahmad Haider •  PhD in “Measurement of energy landscapes of biological interactions using boltzmann sampling” •  Georgia Tech •  Love hiking and reading fiction/non-fiction About Me