Mo'Mentum

01/06/2017 Version of demo slides

Transcript

1. Consulting Proposal September 4, 20XX Mo’Mentum Optimize your mo’mentum for social change Peter Winslow
6. Victory! Ok, let’s do this!
7. Is my petition good enough? Problem: Lots of help for spreading petitions! Not much help for writing them...
8. Solution: Mo’mentum Just what I was looking for! Mo’Mentum Probability of success Time scale to reach signature goal
9. Data Source Change.org sitemap Petition urls Petition id’s Petition text and metadata Change.org api Over ~ 40,000 Petitions
10. Sentiment Feature Engineering Text data Metadata Stopwords, Lemmatization Grammatical Structure Word/Sentence Count Signatures Time Petition momentum= Timestamps Signature count A metric for petition momentum Success/Failure
11. Algorithms: Classification Random Forest Classifier (Scikit-Learn) Predict success/failure of petition 18 features after selection Reasons for choosing: • Lots of complication yet resistant to overfitting Challenges: • Class imbalance in the data Validation: Train-Test-evaluation split with 5-fold CV
12. Algorithms: Regression GradientBoostingRegressor (Scikit-Learn) Predict signature accumulation rate 17 features after selection Reasons for choosing: • Many features, highly non-linear, can return predicted “quantiles” Challenges: • The right evaluation metric? Validation: Train-Test-evaluation split with 5-fold CV
13. About me Peter Winslow The Professional PhD + 1 Postdoc in theoretical High Energy Physics and Cosmology. Specific interests: building models to explain the existence of matter in the Universe + searching for these models in next-generation super colliders! Peter Winslow The New Father! Kiana Winslow, born Nov. 29th 2016