フリー問題

問題文

A team removes the gender column from its loan dataset, retrains, and reports that the model can no longer be unfair because it never sees the attribute. Which source of bias does that argument leave untouched, and which test still applies?

選択肢

1. The loader and the parser can create the unfairness alone, so a known malformed set must be pushed through and the failures across the stage compared.
2. The algorithm and the model can create the unfairness alone, so a known unbiased set must be pushed through and the outcomes across the grouping compared.
3. The cache and the index can create the unfairness alone, so a known cold set must be pushed through and the timings across the shard compared.
4. The console and the report can create the unfairness alone, so a known blank set must be pushed through and the layout across the screen compared.