Machine Learning for Phone-Based Relationship Estimation: The Need to Consider Population ...
summary
Combining phone logs, demographics and labeled places, the study trains automated models and shows adding semantic location boosts accuracy, reaching F1 = 0.68 vs 0.62 when compared to communication features alone. Models are built with autoML for tuning and ensembling, and an age-split experiment finds models trained on younger participants generalize poorly to older groups. The paper provides feature sets, sample-size details, and cross-age performance numbers rather than vague examples.
AI-written summary · about
related talks
21m“Trust Us”: Mobile Phone Use Patterns Can Predict Individual Trust Propensity
29mUser Motivation: Refining Customer Segments with Location
06mUnifying Online and Offline Worlds
09mMachine Learning on the Edge
36mMachine Learning Sucks
40mMachine Learning over Small Data: generating new variables with predictive power
22mPredict the future (or don't) with machine learning in applications
01h 03mUsing Machine Learning and AI to serve people and communities