gistfile1.txt

Moving flagged model files to /mnt/efs_shared/models/, if they exist...
  ...
  copied: /tmp/models/zerocater_pm_bonus__baseline_3_features__44_ratings__8a378ec90200acb1cd93d5dde98408dc
  copied: /tmp/models/was_tech_lead__baseline_3_features__28_ratings__b2fc789a7fe9fa1e36c2409784617770
  copied: /tmp/models/trait__profile_fact__info_amount__baseline_3_features__9739_ratings__8a378ec90200acb1cd93d5dde98408dc
  not found: /tmp/models/trait__personality__technical_ownership__baseline_3_features__2485_ratings__b2fc789a7fe9fa1e36c2409784617770
  ...

Marked 100 models not currently_deployed as should_be_deployed.

Breakdown by experiment_tag:
baseline_3_features    85
random_baseline        15
Name: experiment_tag, dtype: int64
                                               current_score current_tag  \
ability__function__dev_ops                             0.113    baseline   
ability__function__programming                         0.074    baseline   
ability__function__web_back_end_programming            0.103    baseline   
ability__function__web_full_stack_programming          0.232    baseline   
ability__software_tool__javascript                     0.157    baseline   

                                               new_score              new_tag  \
ability__function__dev_ops                         0.353  baseline_3_features   
ability__function__programming                     0.535  baseline_3_features   
ability__function__web_back_end_programming        0.380  baseline_3_features   
ability__function__web_full_stack_programming      0.577  baseline_3_features   
ability__software_tool__javascript                 0.447  baseline_3_features   

                                               improvement  
ability__function__dev_ops                           0.240  
ability__function__programming                       0.461  
ability__function__web_back_end_programming          0.277  
ability__function__web_full_stack_programming        0.345  
ability__software_tool__javascript                   0.290  

Average improvement is 0.221.

Note: I truncated the comparison table results to 5 rows, instead of printing all 100, as DataFrames print awkwardly across multiple lines when a tty isn't wide enough.