fix: correct best_instance_index in batch.select_instance

[zhangyu94]

modAL/modAL/batch.py

Lines 104 to 106 in 452898f

	best_instance_index = np.argmax(scores)
	mask[best_instance_index] = 0
	return best_instance_index, X_pool[best_instance_index].reshape(1, -1), mask

Hi!

This pull request addresses the bug that the computed best_instance_index in batch.select_instance is the index of the best instance in X_pool[mask] instead of the index in X_pool.

For this reason, the returned X_pool[best_instance_index].reshape(1, -1) and mask (updated with mask[best_instance_index] = 0) are both incorrect.

[codecov-io]

Codecov Report

Merging #29 into dev will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##              dev      #29      +/-   ##
==========================================
+ Coverage   98.05%   98.05%   +<.01%     
==========================================
  Files          30       30              
  Lines        1594     1597       +3     
==========================================
+ Hits         1563     1566       +3     
  Misses         31       31

Impacted Files	Coverage Δ
modAL/batch.py	97.72% <100%> (+0.16%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 452898f...f8778f5. Read the comment docs.

[cosmic-cortex]

Thanks! I have reviewed the code, the original interpretation was indeed incorrect in this case.