"Decoding the Urban DNA and Harnes
summary
Sense Networks processes billions of lat/long/timestamp points to build a place-time index that maps where people come from and go after visiting venues, then converts each user’s history into a probabilistic ~500k-length vector matrix that discards raw locations for K-level anonymization. Clustered behavioral graphs identify the top 200 Friday-night spots in San Francisco and produce dynamic demographics by hour. Models power personalization and churn prediction — carrier churn models perform twice as good as baseline — and a deployed Citysense iPhone app shows live aggregate crowds.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:13Where does this come from
- 1:50Other sources of data
- 2:45Can we understand the world
- 3:13Location data
- 3:40Demographics
- 7:20Understanding the Tribes
- 8:14Indexing User Data
- 9:52Dynamic Demographics
- 11:02Turn
- 11:36Conclusion
related talks
15mData-Driven Cities
55mSocial Listening for Richer Data, Unstructured Analytics
28mDiscussion: Humans As Sensors
08mEvil Uses of Data
29mUser Motivation: Refining Customer Segments with Location
26mThe Quantified Self in the Smart City: Geo-Visualizing the Open Data of YOU
14mTelling Traces
56mSocial Listening, Data Extraction & Discovery