226 lines (225 with data), 54.7 kB

{"timestamp": "2025-02-21T17:07:46.381760", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.381982", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.382130", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.382266", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.382390", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.382516", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383090", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383237", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383543", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383580", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383615", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383647", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383678", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383712", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383830", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.383863", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384130", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384230", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384269", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384342", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384374", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384409", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384515", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384547", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384723", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384756", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384798", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384829", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384860", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384893", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.384992", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.385022", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:07:46.385345", "level": "ERROR", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 49, "message": "Error loading data: [Errno 2] No such file or directory: 'data/trials.csv'", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.104679", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.105052", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.105223", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.105360", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.105490", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.105625", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106343", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106487", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106841", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106879", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106916", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.106983", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107019", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107132", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107180", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107381", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107452", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107483", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107516", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107550", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107580", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107613", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107712", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107743", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107888", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107920", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107952", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.107984", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.108016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.108048", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.108156", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.108194", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.139756", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 40, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.143564", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 71, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-21T17:08:24.143659", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:08:27.717636", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.889498", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.889831", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.889991", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.890141", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.890279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.890408", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.890993", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891138", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891458", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891493", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891528", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891571", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891603", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891636", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891745", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891776", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.891944", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892012", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892043", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892107", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892140", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892173", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892268", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892298", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892437", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892469", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892535", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892565", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892596", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892693", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.892724", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.923002", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 40, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.926402", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 71, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-21T17:13:00.926504", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:13:04.299751", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:04.392867", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:13:07.874710", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:13:11.561069", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:13:12.209969", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:12.231156", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:13:12.923518", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:13:13.671478", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:13:14.322011", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:13:14.342275", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.039954", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786300", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "prepare_data", "line": 139, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786756", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 154, "message": "\nPrepared Data Information:", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786800", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 157, "message": "\\nTrain set:", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786831", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 158, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786859", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 159, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786931", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 160, "message": "Number of classes: 5", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786961", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 157, "message": "\\nVal set:", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.786992", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 158, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787019", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 159, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787057", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 160, "message": "Number of classes: 5", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787084", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 157, "message": "\\nTest set:", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787112", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 158, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787138", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 159, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787175", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 160, "message": "Number of classes: 5", "taskName": null}
{"timestamp": "2025-02-21T17:13:15.787202", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 163, "message": "\\nFeature types used: ['tfidf', 'statistics']", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.398461", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.398883", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.399044", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.399177", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.399308", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.399429", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400065", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400191", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400485", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400518", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400550", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400580", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400608", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400639", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400744", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400773", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.400941", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401001", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401034", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401064", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401092", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401119", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401152", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401248", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401407", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401437", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401466", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401498", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401525", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401553", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401644", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:22:34.401673", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.575175", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.575522", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.575701", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.575862", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.576005", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.576131", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.576721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.576876", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577223", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577258", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577293", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577325", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577357", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577391", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577506", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577537", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577707", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577771", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577808", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577839", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577868", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577897", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.577930", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578028", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578060", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578199", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578230", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578261", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578290", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578322", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578352", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578446", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.578475", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.608714", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.611124", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-21T17:26:46.611213", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:26:49.977749", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:50.067802", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:26:53.529270", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:26:57.221409", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:26:57.870788", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:26:57.892239", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:26:58.607210", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:26:59.354215", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-21T17:27:00.005056", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-21T17:27:00.025778", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-21T17:27:00.717760", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.460585", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.465474", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.471486", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472383", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472448", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472490", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472626", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472658", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472687", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472813", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472843", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-21T17:27:01.472873", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}