de presentation

17
SceneFindr Stephanie Stark

Upload: scstark

Post on 22-Jan-2018

181 views

Category:

Data & Analytics


0 download

TRANSCRIPT

SceneFindrStephanie Stark

Motivation

● Interested in hearing live music, but don’t know where to go?

Pipeline

Data Sources

Data Sources

Data Sources

Data Sources

Data Sources

Pipeline

ETL

Artists

Events

Feature Extraction

K-Means Clustering

Recommendations

Database

Pipeline

Lessons Learned (the hard way!)

● Scala● Parallelized ML algorithms

About Me

B.A., Mount Holyoke CollegeMajor: MathematicsMinor: Computer Science

Education

Interests ReadingArt HistoryHiking

Stephanie Stark

Future Work

Implement TF/IDF compatibility for projectUse PCAImplement cosine similarity for feature clusteringCluster within metro areaUse Redis as a cache for feature vectors

Scaling

500GB of artist data500GB of event data