A Tutorial on Matching-based Causal Analysis of Human Behavior Using Smartphone Sensor Data

This repository includes:

• Preprocessing Code of GPS, App Usage, and Activity

  • GPS was categorized into HOME, WORK, OTHERS, and MOVE
  • Application was categorized into SOCIAL, ENTER, HEALTH, WORK, SYSTEM
  • Activity predicted by Android API: STILL, WALKING, RUNNING, IN_VEHICLE, ON_BICYCLE

• Preprocessed Dataset used for the causal inference.

  • 3 types of data were all preprocessed into 15 Minute Interval.
    • Launch Count, Usage Time of each app category
    • Duration of still(DUT_SEDENTARY) and Location(DUT_HOME, DUT_WORK, DUT_OTHER)

• Causal Inference Code for Preprocessed Dataset

  • The code is written in R.