updated models for 2025

retrained models using games published through 2022
moved pipeline into Google Cloud Storage

hurdle

added hurdle model

alpha-lightgbm: Dev/lightgbm (#27)

shifted model types to lightgbm

functional pipeline + reports

increased functionality within targets pipeline for additional modeling/testing
added quarto reports for GitHub pages

0.1.0

summary

using penalized regressions (glmnet) to predict bgg outcomes

description

targets pipeline trains and evaluates models for averageweight, average, and usersrated, then combines these to estimate the geek rating (bayesaverage)

evaluation

models trained through 2019; evaluated on 2020-22