score-nd float rounding error

[Jacob-Le]

When attempting to apply score-nd to a particular file using version 1.3.2, I get the following error (parameters and steps to reproduce to follow):

File "/Users/e35003/Documents/gitclones/ccu_maksimus/ccu_validation_scoring/CCU_validation_scoring/score_submission.py", line 142, in score_nd_submission_dir_cli
score_tad(ref, merged_hyp, "norm", thresholds, args.output_dir, mapping_df, float(args.time_span_scale_collar), float(args.text_span_scale_collar), args.align_hacks)
File "/Users/e35003/Documents/gitclones/ccu_maksimus/ccu_validation_scoring/CCU_validation_scoring/score_norm_emotion.py", line 778, in score_tad
sumup_tad_class_level_scores(pr_iou_scores, iou_thresholds, output_dir, class_type)
File "/Users/e35003/Documents/gitclones/ccu_maksimus/ccu_validation_scoring/CCU_validation_scoring/score_norm_emotion.py", line 697, in sumup_tad_class_level_scores
agg_table['value'] = agg_table['value'].round(3)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/e35003/Documents/virtual_environments/MAKSIMUS/lib/python3.11/site-packages/pandas/core/series.py", line 2688, in round
result = self._values.round(decimals)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: loop of ufunc does not support argument 0 of type float which has no callable rint method

---

Python version: 3.11
Pandas version: 2.1.3
Numpy version: 1.26.2
Scipy version: 1.11.4

Parameters used:

CCU_scoring score-nd -s <path_to_submission> -ref <path_to_ldc2023e21_version_4.0> -i <path_to_scoring_index> -o <output_path> -t intersection:gt:0 -aR 30 -xR 300 -vR class -aC 15 -xC 150

---

Scoring index file:

file_id
N01006WA4

---

Isolated submission file that causes the error (file has been included as a csv, simply need to change the extension to .tab N01006WA4.csv):

file_id norm start end status llr
N01006WA4 102 104.84 123.4 adhere 0.4697356
N01006WA4 502 10.89 17.49 adhere 0.2709458
N01006WA4 108 24.07 68.22 adhere 0.4048885
N01006WA4 102 135.44 144.52 adhere 0.5015393
N01006WA4 109 165.0 180.0 adhere 0.5656774

---

Example of a submission file where the error does not occur N01005FR3.csv:

file_id norm start end status llr
N01005FR3 109 60.0 90.0 adhere 0.653696
N01005FR3 109 225.0 255.0 adhere 0.5354934
N01005FR3 109 300.0 315.0 adhere 0.6340112
N01005FR3 106 0.0 30.0 adhere 0.9711443
N01005FR3 103 1.88 7.46 adhere 0.2993998
N01005FR3 101 85.09 88.93 adhere 0.0919553
N01005FR3 109 165.0 180.0 violate 0.7992395
N01005FR3 109 375.0 390.0 adhere 0.6774845

[Jacob-Le]

For reference, in an email conversation the issue was that the pandas version for this release needed to be 1.5.3, but was instead 2.1.3. I was advised that the issue would be solved in a future release