Zeitpunkt Nutzer Delta Tröts TNR Titel Version maxTL Di 10.09.2024 00:00:27 2.878 0 690.848 240,0 digitalcourage.social 4.2.12 1.024 Mo 09.09.2024 00:01:16 2.878 -3 698.523 242,7 digitalcourage.social 4.2.12 1.024 So 08.09.2024 00:01:04 2.881 -1 699.213 242,7 digitalcourage.social 4.2.12 1.024 Sa 07.09.2024 00:01:02 2.882 0 698.644 242,4 digitalcourage.social 4.2.12 1.024 Fr 06.09.2024 00:00:28 2.882 0 698.046 242,2 digitalcourage.social 4.2.12 1.024 Do 05.09.2024 00:00:53 2.882 +2 697.416 242,0 digitalcourage.social 4.2.12 1.024 Mi 04.09.2024 00:01:03 2.880 +1 699.551 242,9 digitalcourage.social 4.2.12 1.024 Di 03.09.2024 00:01:04 2.879 -1 698.658 242,7 digitalcourage.social 4.2.12 1.024 Mo 02.09.2024 00:01:10 2.880 0 697.817 242,3 digitalcourage.social 4.2.12 1.024 So 01.09.2024 00:01:09 2.880 0 696.680 241,9 digitalcourage.social 4.2.12 1.024
Ralf Stubner (@rstub) · 11/2022 · Tröts: 6.924 · Folger: 267
Di 10.09.2024 00:43
After reading a blog post by @tslumley I was surprised to learn that weighted sampling w/o replacement can mean that the probability of inclusion is not proportional to the weight. Another blog post by @peter_ellis brought some additional explanations and visualizations. Now I have done my own initial comparisons also looking at performance of various methods:
https://stubner.me/post/2024-09-09-algorithms-for-weighted-sampling-without-replacement-part-1/
[Öffentlich] Antw.: 1 Wtrl.: 2 Fav.: 2 · via Mona for iPhone