A lightweight R script that leverages a past dating history and current event log to estimate when one is likely to find a meaningful match (high specificity) in Östergötland, Sweden. It fits a three‐stage Bayesian logistic model (meet, follow-up, resonate), uses informative beta priors (with fallback to empirical priors), and simulates the waiting‐time distribution for the "next relationship". Returns key quantiles (5th - 95th) along with a deterministic worst‐case scenario.
-
Event logging: record venue, crowd size, vibe (1, cold, to 4, warm), and micro-signals (e.g. recognition or unsolicited contact).
-
Bayesian modeling: three conditional Bernoulli regressions (meet, follow-up, resonate) with splines on date or linear time, fit via
brms. -
Informative priors: beta-prior fallback when data are sparse (stage 2 & stage 3), plus Student t-priors on coefficients.
-
Horizon simulation: Monte Carlo samples of waiting events to months; outputs median, quartiles, and tail quantiles.
-
Worst-case deterministic: uses fixed low probability (p = 0.009) to show an absolute lower bound on the waiting horizon.
Linn Friberg