Research talk:Measuring the effect of cross-linking missing articles in English Wikivoyage/Work log/2019-06-03

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

Monday, June 3, 2019[edit]

I've been bad at making log entries. I've been experimenting with timeseries methods. I finally have a working forecast model based on weekly article creations.

See https://paws-public.wmflabs.org/paws-public/User:EpochFail/enwikivoyage-creations/R/weekly_timeseries_analysis.ipynb

I learned from this model that anonymous article creations are still in-line with the model's forecast.

One Sample t-test

data:  d_anon_with_forecast[!is.na(diff), ]$diff
t = -0.437, df = 7, p-value = 0.3376
alternative hypothesis: true mean is less than 0
95 percent confidence interval:
     -Inf 4.598506
sample estimates:
mean of x 
-1.378704

This means we're getting about 1.3 fewer anon article creations than we expect per week over the last few weeks, but the effect is too small to be sure it's not just statistical noise. --Halfak (WMF) (talk) 21:58, 3 June 2019 (UTC)Reply[reply]