Add logger module, and adding wrapper logs to run scripts, will add d…

…eeper level logs in next commit

examples/run_scripts_v6/temp_script_v6.py

@@ -2,45 +2,58 @@
 
 import pandas as pd
 
+from tabrepo.scripts_v6.logging_config import setup_logger
 from tabrepo import load_repository, EvaluationRepository
 from tabrepo.scripts_v6.TabPFN_class import CustomTabPFN
 from tabrepo.scripts_v6.TabPFNv2_class import CustomTabPFNv2
 from tabrepo.scripts_v6.LGBM_class import CustomLGBM
 from experiment_utils import run_experiments, convert_leaderboard_to_configs
 
+logger = setup_logger(log_file_name='temp_script_v6')
+
 if __name__ == '__main__':
-    # Load Context
+
+    logger.info("Starting execution script...")
+
     context_name = "D244_F3_C1530_30"
-    repo: EvaluationRepository = load_repository(context_name, cache=True)
+    logger.info(f"Loading repository for context: {context_name}")
+    try:
+        repo: EvaluationRepository = load_repository(context_name, cache=True)
+        logger.info("Repository loaded successfully.")
+    except Exception as e:
+        logger.error(f"Failed to load repository: {e}", exc_info=True)
+        raise
 
     expname = "./initial_experiment_tabpfn_v6"  # folder location of all experiment artifacts
     ignore_cache = True  # set to True to overwrite existing caches and re-run experiments from scratch
 
     # To run everything:
     # datasets = repo.datasets
     # folds = repo.folds
-
+    folds = [0]
     datasets = [
         "blood-transfusion-service-center",  # binary
         "Australian",  # binary
         "balance-scale",  # multiclass
         # "MIP-2016-regression",  # regression
     ]
+    logger.info(f"Selected Datasets: {datasets}")
+    logger.info(f"Folds to run: {folds}")
 
-    folds = [0]
-
-    # Add a check here if the dataset belong to repo
-    tids = [repo.dataset_to_tid(dataset) for dataset in datasets]
+    try:
+        tids = [repo.dataset_to_tid(dataset) for dataset in datasets]
+    except Exception as e:
+        logger.warning(f"Some datasets may not belong to the repository: {e}", exc_info=True)
 
     methods_dict = {
         "LightGBM": {
             "learning_rate": 0.15,
             "num_leaves": 32,
+            "verbose": -1,  # To suppress warnings
         },
         "TabPFN": {
             "device": 'cpu',
             "N_ensemble_configurations": 32,
-            "max_num_features": 50,
         },
     }
     method_cls_dict = {
@@ -49,20 +62,32 @@
         "TabPFNv2": CustomTabPFNv2,
     }
     methods = list(methods_dict.keys())
+    logger.info(f"Methods to run: {methods}")
 
-    results_lst = run_experiments(
-        expname=expname,
-        tids=tids,
-        folds=repo.folds,
-        methods=methods,
-        methods_dict=methods_dict,
-        method_cls=method_cls_dict,
-        task_metadata=repo.task_metadata,
-        ignore_cache=ignore_cache,
-    )
+    logger.info("Running experiments...")
+    try:
+        results_lst = run_experiments(
+            expname=expname,
+            tids=tids,
+            folds=folds,
+            methods=methods,
+            methods_dict=methods_dict,
+            method_cls=method_cls_dict,
+            task_metadata=repo.task_metadata,
+            ignore_cache=ignore_cache,
+        )
+        logger.info("Experiments Status: Successful.")
+    except Exception as e:
+        logger.error(f"An error occurred while running experiments: {e}", exc_info=True)
+        raise
 
-    results_df = pd.concat(results_lst, ignore_index=True)
+    logger.info("Concatenating results into Dataframe...")
+    try:
+        results_df = pd.concat(results_lst, ignore_index=True)
+    except Exception as e:
+        logger.error(f"An error occurred while concatenating results: {e}", exc_info=True)
 
+    logger.info("Renaming leaderboard columns... ")
     results_df = convert_leaderboard_to_configs(results_df)
     with pd.option_context('display.max_rows', None, 'display.max_columns', None, 'display.width', 1000):
         print(results_df)
@@ -77,21 +102,34 @@
         "NeuralNetTorch_c1_BAG_L1",
         "NeuralNetFastAI_c1_BAG_L1",
     ]
+    logger.info(f"Comparison configs: {comparison_configs}")
 
     baselines = [
         "AutoGluon_bq_4h8c_2023_11_14",
     ]
+    logger.info(f"Baseline: {baselines}")
+
+    logger.info(f"Comparing metrics...")
+    try:
+        metrics = repo.compare_metrics(
+            results_df,
+            datasets=datasets,
+            folds=repo.folds,
+            baselines=baselines,
+            configs=comparison_configs,
+        )
+    except Exception as e:
+        logger.error(f"An error occurred in compare_metrics(): {e}", exc_info=True)
+        raise
 
-    metrics = repo.compare_metrics(
-        results_df,
-        datasets=datasets,
-        folds=repo.folds,
-        baselines=baselines,
-        configs=comparison_configs,
-    )
     with pd.option_context("display.max_rows", None, "display.max_columns", None, "display.width", 1000):
         print(f"Config Metrics Example:\n{metrics}")
-    evaluator_output = repo.plot_overall_rank_comparison(
-        results_df=metrics,
-        save_dir=expname,
-    )
+
+    logger.info("Plotting overall rank comparison...")
+    try:
+        evaluator_output = repo.plot_overall_rank_comparison(
+            results_df=metrics,
+            save_dir=expname,
+        )
+    except Exception as e:
+        logger.error(f"An error occurred in plot_overall_rank_comparison(): {e}", exc_info=True)

tabrepo/scripts_v6/logging_config.py

@@ -0,0 +1,32 @@
+"""
+**logger** module just exposes a ``setup_logger`` function to quickly configure the python logger.
+"""
+import logging
+import os
+
+
+# WIP - Prateek: TODO: pickup filenames dynamically
+# TODO: Improve log_dir path
+def setup_logger(log_file_name, level=logging.INFO):
+    """Set up a logger with a specific log file name."""
+    logger = logging.getLogger(log_file_name)
+
+    if not logger.hasHandlers():
+        logger.setLevel(level)
+
+        log_dir = os.path.join(os.path.dirname(__file__), '..', '..', 'logs')
+        os.makedirs(log_dir, exist_ok=True)
+
+        log_file = os.path.join(log_dir, log_file_name)
+        file_handler = logging.FileHandler(log_file)
+
+        stream_handler = logging.StreamHandler()
+
+        formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+        file_handler.setFormatter(formatter)
+        stream_handler.setFormatter(formatter)
+
+        logger.addHandler(file_handler)
+        logger.addHandler(stream_handler)
+
+    return logger