yeatmanlab / jsCAT

100%

main: 100%

Pull #35

github

Prior change (#38)

* Add simple test for EAP. It fails

* fix: correct EAP ability estimation by exponentiating log-likelihood

The EAP ability estimation was incorrectly using log-likelihood values
directly in probability calculations. This fix:

1. Converts log-likelihood back to probability space using Math.exp()
   before using it in the EAP calculation
2. Updates test expectations to match the corrected behavior
3. Simplifies the EAP test to only check that theta decreases
   after an incorrect response to an easy item

The EAP method now properly reduces ability estimates when examinees
fail easy items, which is critical for accurate adaptive testing.

* update test and validation code

* add uniform prior

* update lint fix

* update

* update

* Update src/utils.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/utils.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/__tests__/cat.test.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/__tests__/cat.test.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* update

* update validateprior as a return function

* Update src/cat.ts

Co-authored-by: Adam Richie-Halford <richford@users.noreply.github.com>

* Update cat.test.ts

---------

Co-authored-by: Adam Richie-Halford <richiehalford@gmail.com>
Co-a... (continued)

318 of 318 branches covered (100.0%)

Branch coverage included in aggregate %.

88 of 88 new or added lines in 2 files covered. (100.0%)

1504 of 1504 relevant lines covered (100.0%)

19998.22 hits per line