to read, I respect cuicui a lot.
| description | Most statistical and ML models are built to predict an observable truth. Sometimes, however, we need to be able to infer abstract constructs such as risk tolerance, user trust, brand affinity or programming skill. This article explores how to extract these latent variables from noisy proxies using **Item Response Theory**. |
| tags | clippings |
| source | https://cuicui.dev/en/blog/modeling-the-unobservable/ |
1 min read
to read, I respect cuicui a lot.