Inference on ordered categorical data via ordered categorical GLM's
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

This work builds a Bayesian ordered categorical linear model to estimate the ordered multinomial distribution underlying product feedback scores. "I'm testing website variants and have collected 50 feedback scores on the scale 1-5: what's the true distribution of scores look like?" in effect.

The model is then compared to a vanilla multinomial regression on the same data. In addition, it includes a Shiny app which A/B tests two distinct populations of ordered integer scores. An example of its output is given below.

Finally, this work is accompanied by a blog post which can be found at the following link.

Shinyapps output example