This PR introduces a new Test method, chi_square_sc(level=0.05, return_diagnostics=False). It is required to produce the Quantum-like tests against total, that follow a completely different approach than Dimensions (and Askia).
I have modified the respective code in the main test algorithm to combine the new Chi^2-test results with the established output computation. I am doing this by skipping the "regular" test against total results and instead append the Chi^2 ones.
Some notes:
I have also modified the view_maps to always force the total test for Link, i.e. it's enough to instruct the tests as per
b.set_sigtests(levels=0.05, mimic='LINK_legacy') in NB5; NB6 and NB8 can be run as usual.
Test results are indicated as before using the green-up & red-down arrows.
Chi^2 tests are not performed on means.
Documentation can be found in our Confluence pages. George can also be consulted for additional testing vs. Quantum deliverables.
This PR introduces a new
Test
method,chi_square_sc(level=0.05, return_diagnostics=False)
. It is required to produce the Quantum-like tests against total, that follow a completely different approach than Dimensions (and Askia).I have modified the respective code in the main test algorithm to combine the new Chi^2-test results with the established output computation. I am doing this by skipping the "regular" test against total results and instead append the Chi^2 ones.
Some notes:
view_maps
to always force the total test for Link, i.e. it's enough to instruct the tests as perb.set_sigtests(levels=0.05, mimic='LINK_legacy')
in NB5; NB6 and NB8 can be run as usual.Documentation can be found in our Confluence pages. George can also be consulted for additional testing vs. Quantum deliverables.
Unit tests are passing (skipping Excel ones):