This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2020-10-08
Channels
- # announcements (10)
- # babashka (4)
- # beginners (98)
- # cider (47)
- # clara (6)
- # clj-kondo (16)
- # clojure (54)
- # clojure-australia (3)
- # clojure-berlin (3)
- # clojure-czech (2)
- # clojure-europe (77)
- # clojure-nl (4)
- # clojure-uk (12)
- # clojuredesign-podcast (6)
- # clojurescript (10)
- # conjure (56)
- # cursive (3)
- # data-science (6)
- # datascript (8)
- # datomic (213)
- # depstar (5)
- # events (1)
- # figwheel-main (2)
- # fulcro (23)
- # graalvm (2)
- # jobs (3)
- # london-clojurians (1)
- # malli (30)
- # meander (15)
- # midje (1)
- # mount (5)
- # off-topic (18)
- # re-frame (4)
- # reitit (15)
- # remote-jobs (1)
- # shadow-cljs (23)
- # spacemacs (10)
- # specter (1)
- # tools-deps (88)
- # vim (16)
- # xtdb (1)
I'm a noob in statistics but have been watching some videos here: https://creativemaths.net/videos/#inference As a learning excercise, I tried to implement confidence interval calculations for mean and median here: https://github.com/jumarko/clojure-experiments/blob/master/src/clojure_experiments/stats/confidence_intervals.clj#L37-L92 Does the implementation makes any sense? Any good clojure/java libraries that do the same thing? (especially interested in approaches for calculating a confidence interval for a median)
Compare with my implementation here: https://github.com/generateme/fastmath/blob/382ec27d23fbaf8925a088c5d5edc27e4596dd0d/src/fastmath/stats.clj#L407
There are also other intervals/extents like adjacent values or median/quantiles etc.
There is also bootstrap ci https://github.com/generateme/fastmath/blob/382ec27d23fbaf8925a088c5d5edc27e4596dd0d/src/fastmath/stats.clj#L423
@U06BE1L6T your clojure_experiments
repo looks amazing. What a huge collection of interesting examples! ✨
As @U1EP3BZ3Qsuggested, I think Bootstrap-based confidence intervals seem like a good practical approach for confidence intervals for medians, as there isn't any general approximation such as the t
-distribution we have for means.
The intro to bootstrap that is linked to from @U1EP3BZ3Q's code is quite nice to read, I think.