Named Entity Recognition Testing Config
Below is the default configuration for all Named Entity Recognition tests. A copy of this can also be found in your rime_trial bundle (inside the nlp_examples/ner/default_test_config.json).
{
"categories": [],
"run_default": null,
"custom_tests": null,
"numeric_outlier": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
},
"min_normal_prop": 0.99,
"baseline_quantile": 0.1,
"perturb_multiplier": 1.0
},
"unseen_categorical": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"unseen_domain": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"unseen_email": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"unseen_url": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"rare_categories": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
},
"include_columns": [],
"min_num_occurrences": 0,
"min_pct_occurrences": 0,
"min_ratio_rel_uniform": 0.005
},
"out_of_range": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
},
"std_factor": 3
},
"int_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"float_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"str_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"bool_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"url_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"domain_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"email_feature_type": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"req_characters": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
},
"column_specific_params": {}
},
"inconsistencies": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
},
"freq_ratio_threshold": 0.02,
"min_correlation": 0.1,
"max_pairwise_tests": 200,
"max_unique_pairs_for_firewall": 15
},
"null_check": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"capitalization": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"empty_string": {
"exclude_columns": [],
"run": true,
"performance_change_config": {
"num_samples_to_simulate": 100,
"min_num_samples": 10,
"ignore_errors": false,
"severity_thresholds": null
}
},
"feat_subset_auc": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_accuracy": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_f1": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_macro_f1": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_precision": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_macro_precision": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_fpr": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_recall": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_macro_recall": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_pred_variance_pos": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_pred_variance_neg": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_pred_variance_all": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_rmse": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_mae": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_rank_correlation": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_ndcg": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"feat_subset_mrr": {
"exclude_columns": [],
"run": true,
"min_sample_size": 20,
"performance_change_thresholds": null
},
"correlation_drift": {
"exclude_columns": [],
"run": true,
"min_correlation": 0.1,
"correlation_thresholds": [
0.1,
0.2,
0.3
],
"p_value_threshold": 0.05,
"max_pairwise_tests": 200
},
"mutual_information_feat_drift": {
"exclude_columns": [],
"run": true,
"min_mutual_information": 0.1,
"mutual_information_thresholds": [
0.1,
0.2,
0.3
],
"max_pairwise_tests": 200,
"min_sample_size": 100
},
"mutual_information_label_drift": {
"exclude_columns": [],
"run": true,
"min_mutual_information": 0.1,
"mutual_information_thresholds": [
0.1,
0.2,
0.3
],
"max_pairwise_tests": 200,
"min_sample_size": 100
},
"categorical_label_drift": {
"run": true,
"drift_statistic": "Population Stability Index",
"params": {
"run": true,
"num_values_for_graph": 5,
"distance_thresholds": [
0.2,
0.4,
0.6
]
}
},
"multiclass_pred_label_drift": {
"run": true,
"drift_statistic": "Population Stability Index",
"params": {
"run": true,
"num_values_for_graph": 5,
"distance_thresholds": [
0.2,
0.4,
0.6
]
}
},
"regression_label_drift": {
"run": true,
"p_value_threshold": 0.05,
"ks_stat_thresholds": [
0.1,
0.33,
0.67
]
},
"categorical_drift": {
"exclude_columns": [],
"run": true,
"drift_statistic": "Population Stability Index",
"params": {
"run": true,
"drift_scaling_factor": 0.005,
"performance_change_thresholds": null,
"min_sample_size": 100,
"max_sample_size": null,
"distance_threshold": 0.2
}
},
"null_proportion": {
"exclude_columns": [],
"run": true,
"drift_scaling_factor": 0.005,
"performance_change_thresholds": null,
"p_value_threshold": 0.05,
"min_sample_size": 100
},
"row_null_proportion": {
"exclude_columns": [],
"run": true,
"drift_statistic": "Population Stability Index",
"params": {
"exclude_columns": [],
"run": true,
"drift_scaling_factor": 0.005,
"performance_change_thresholds": null,
"psi_threshold": 0.2
}
},
"continuous_drift": {
"exclude_columns": [],
"run": true,
"drift_scaling_factor": 0.005,
"performance_change_thresholds": null,
"drift_statistic": "Population Stability Index",
"params": {
"run": true,
"drift_scaling_factor": 0.005,
"performance_change_thresholds": null,
"min_sample_size": 100,
"min_num_quantiles": 1000,
"distance_threshold": 0.2,
"num_bins": 100
}
},
"overall_metrics": {
"run": true,
"metrics_specific_thresholds": {}
},
"prediction_drift": {
"run": true,
"drift_statistic": "Population Stability Index",
"params": {
"run": true,
"min_sample_size": 100,
"min_num_quantiles": 1000,
"psi_thresholds": [
0.2,
0.4,
0.6
],
"num_bins": 100
}
},
"calibration_comparison": {
"run": true,
"severity_level_thresholds": [
0.02,
0.06,
0.1
]
},
"global_sample_size": null,
"metadata_tests": null,
"unseen_unigram_abnormal_input": {
"run": true,
"severity_thresholds": [
0.0,
0.05,
0.1
],
"performance_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"empty_string_abnormal_input": {
"run": true,
"severity_thresholds": [
0.0,
0.05,
0.1
],
"performance_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"char_dist_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 0,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"unigrams_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 5,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"bigrams_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 5,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"lower_case_entity": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"upper_case_entity": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"ampersand": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"abbreviation": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"white_space_special_chars": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"ascii": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"remove_special_chars": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"swap_seen_entities": {
"run": true,
"sample_size": 200,
"severity_thresholds": [
0,
0.1,
0.3
]
},
"num_entities_subset": {
"run": true,
"metric_name": null,
"num_subsets": 5,
"min_subset_size": 10,
"severity_thresholds": [
0.05,
0.08,
0.13
]
},
"ground_truth_entity_type": {
"run": true,
"metric_name": null,
"num_subsets": 5,
"min_subset_size": 10,
"severity_thresholds": [
0.05,
0.08,
0.13
]
},
"predicted_entity_type": {
"run": true,
"metric_name": null,
"num_subsets": 5,
"min_subset_size": 10,
"severity_thresholds": [
0.05,
0.08,
0.13
]
},
"entity_types_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 0,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"predicted_entity_types_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 0,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
}
},
"entity_lengths_drift": {
"run": true,
"drift_metrics": [
{
"distance_metric": "Population Stability Index",
"severity_threshold": [
0.1,
0.2,
0.4
]
}
],
"min_occurrences": 0,
"model_impact_config": {
"ignore_observed_performance": false,
"min_num_samples": 10,
"severity_thresholds": [
0.01,
0.05,
0.1
]
},
"num_quantiles": 1001,
"num_bins": 100
}
}