<?xml version="1.0" encoding="utf-8" ?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:r="https://r-universe.dev"><channel><title>veronicazwt.r-universe.dev</title><link>https://veronicazwt.r-universe.dev</link><description>Recent package updates in veronicazwt</description><generator>R-universe</generator><image><url>https://github.com/veronicazwt.png</url><title>R packages by veronicazwt</title><link>https://veronicazwt.r-universe.dev</link></image><lastBuildDate>Sat, 22 Aug 2026 00:16:30 GMT</lastBuildDate><item><title>[kanji709] marp 0.1.1</title><author>wantungtsou@gmail.com (Veronica Tsou)</author><description>To implement a model-averaging approach with different
renewal models, with a primary focus on forecasting large
earthquakes. Based on six renewal models (i.e., Poisson, Gamma,
Log-Logistics, Weibull, Log-Normal and BPT), model-averaged
point estimates are calculated using AIC weights. Additionally,
both percentile and studentized bootstrapped model-averaged
confidence intervals are constructed. In comparison, point and
interval estimation from the individual or &quot;best&quot; model
(determined via model selection) can be retrieved.</description><link>https://github.com/r-universe/kanji709/actions/runs/35584678120</link><pubDate>Sat, 22 Aug 2026 00:16:30 GMT</pubDate><r:package>marp</r:package><r:version>0.1.1</r:version><r:status>success</r:status><r:repository>https://kanji709.r-universe.dev</r:repository><r:upstream>https://github.com/kanji709/marp</r:upstream><r:article><r:source>marp-workflow.Rmd</r:source><r:filename>marp-workflow.html</r:filename><r:title>A workflow for model-averaged renewal-process inference with marp</r:title><r:created>2026-08-21 01:57:15</r:created><r:modified>2026-08-22 00:16:30</r:modified></r:article></item></channel></rss>