halib 0.1.65__py3-none-any.whl → 0.1.67__py3-none-any.whl

halib/research/metrics.py

@@ -0,0 +1,134 @@
+# -------------------------------
+# Metrics Backend Interface
+# -------------------------------
+import inspect
+from typing import Dict, Union, List, Any
+from abc import ABC, abstractmethod
+
+class MetricsBackend(ABC):
+    """Interface for pluggable metrics computation backends."""
+
+    def __init__(self, metrics_info: Union[List[str], Dict[str, Any]]):
+        """
+        Initialize the backend with optional metrics_info.
+        """
+        self.metric_info = metrics_info
+        self.validate_metrics_info(self.metric_info)
+
+    @property
+    def metric_names(self) -> List[str]:
+        """
+        Return a list of metric names.
+        If metric_info is a dict, return its keys; if it's a list, return it directly.
+        """
+        if isinstance(self.metric_info, dict):
+            return list(self.metric_info.keys())
+        elif isinstance(self.metric_info, list):
+            return self.metric_info
+        else:
+            raise TypeError("metric_info must be a list or a dict")
+
+    def validate_metrics_info(self, metrics_info):
+        if isinstance(metrics_info, list):
+            return metrics_info
+        elif isinstance(metrics_info, dict):
+            return {k: v for k, v in metrics_info.items() if isinstance(k, str)}
+        else:
+            raise TypeError(
+                "metrics_info must be a list of strings or a dict with string keys"
+            )
+
+    @abstractmethod
+    def compute_metrics(
+        self, metrics_info: Union[List[str], Dict[str, Any]],  metrics_data_dict: Dict[str, Any], *args, **kwargs
+    ) -> Dict[str, Any]:
+        pass
+
+    def prepare_metrics_backend_data(
+        self, raw_metric_data, *args, **kwargs
+    ):
+        """
+        Prepare the data for the metrics backend.
+        This method can be overridden by subclasses to customize data preparation.
+        """
+        return raw_metric_data
+
+    def calc_metrics(
+        self,  metrics_data_dict: Dict[str, Any], *args, **kwargs
+    ) -> Dict[str, Any]:
+        """
+        Calculate metrics based on the provided metrics_info and data.
+        This method should be overridden by subclasses to implement specific metric calculations.
+        """
+        # prevalidate the metrics_data_dict
+        for metric in self.metric_names:
+            if metric not in metrics_data_dict:
+                raise ValueError(f"Metric '{metric}' not found in provided data.")
+        # Prepare the data for the backend
+        metrics_data_dict = self.prepare_metrics_backend_data(
+            metrics_data_dict, *args, **kwargs
+        )
+        # Call the abstract method to compute metrics
+        return self.compute_metrics(self.metric_info, metrics_data_dict, *args, **kwargs)
+
+class TorchMetricsBackend(MetricsBackend):
+    """TorchMetrics-based backend implementation."""
+
+    def __init__(self, metrics_info: Union[List[str], Dict[str, Any]]):
+        try:
+            import torch
+            from torchmetrics import Metric
+        except ImportError:
+            raise ImportError(
+                "TorchMetricsBackend requires torch and torchmetrics to be installed."
+            )
+        self.metric_info = metrics_info
+        self.torch = torch
+        self.Metric = Metric
+        self.validate_metrics_info(metrics_info)
+
+    def validate_metrics_info(self, metrics_info):
+        if not isinstance(metrics_info, dict):
+            raise TypeError(
+                "TorchMetricsBackend requires metrics_info as a dict {name: MetricInstance}"
+            )
+        for k, v in metrics_info.items():
+            if not isinstance(k, str):
+                raise TypeError(f"Key '{k}' is not a string")
+            if not isinstance(v, self.Metric):
+                raise TypeError(f"Value for key '{k}' must be a torchmetrics.Metric")
+        return metrics_info
+
+    def compute_metrics(self, metrics_info, metrics_data_dict, *args, **kwargs):
+        out_dict = {}
+        for metric, metric_instance in metrics_info.items():
+            if metric not in metrics_data_dict:
+                raise ValueError(f"Metric '{metric}' not found in provided data.")
+
+            metric_data = metrics_data_dict[metric]
+            sig = inspect.signature(metric_instance.update)
+            expected_args = list(sig.parameters.values())
+
+            if isinstance(metric_data, dict):
+                args = [metric_data[param.name] for param in expected_args]
+            elif isinstance(metric_data, (list, tuple)):
+                args = metric_data
+            else:
+                raise TypeError(f"Unsupported data format for metric '{metric}'")
+
+            if len(expected_args) == 1:
+                metric_instance.update(args)
+            else:
+                metric_instance.update(*args)
+
+            computed_value = metric_instance.compute()
+            if isinstance(computed_value, self.torch.Tensor):
+                computed_value = (
+                    computed_value.item()
+                    if computed_value.numel() == 1
+                    else computed_value.tolist()
+                )
+
+
+            out_dict[metric] = computed_value
+        return out_dict

halib/research/perfcalc.py

@@ -1,71 +1,27 @@
 import os
 import glob
-import inspect
+from typing import Optional, Tuple
 import pandas as pd
 
-from typing import Dict
-from functools import wraps
 from rich.pretty import pprint
 
 from abc import ABC, abstractmethod
+from collections import OrderedDict
 
 from ..filetype import csvfile
+from ..system import filesys as fs
 from ..common import now_str
 from ..research.perftb import PerfTB
-from collections import OrderedDict
+from ..research.metrics import *
 
-# try to import torch, and torchmetrics
-try:
-    import torch
-    import torchmetrics
-    from torchmetrics import Metric
-except ImportError:
-    raise ImportError("Please install torch and torchmetrics to use this module.")
-
-def validate_torch_metrics(fn):
-    @wraps(fn)
-    def wrapper(self, *args, **kwargs):
-        result = fn(self, *args, **kwargs)
-
-        if not isinstance(result, dict):
-            raise TypeError("torch_metrics() must return a dictionary")
-
-        for k, v in result.items():
-            if not isinstance(k, str):
-                raise TypeError(f"Key '{k}' is not a string")
-            if not isinstance(v, Metric):
-                raise TypeError(
-                    f"Value for key '{k}' is not a torchmetrics.Metric (got {type(v).__name__})"
-                )
-
-        return result
-
-    return wrapper
-def valid_custom_fields(fn):
-    @wraps(fn)
-    def wrapper(self, *args, **kwargs):
-        rs = fn(self, *args, **kwargs)
-        if not isinstance(rs, tuple) or len(rs) != 2:
-            raise ValueError("Function must return a tuple (outdict, custom_fields)")
-        outdict, custom_fields = rs
-        if not isinstance(outdict, dict):
-            raise TypeError("Output must be a dictionary")
-        if not isinstance(custom_fields, list):
-            raise TypeError("Custom fields must be a list")
-        for field in custom_fields:
-            if not isinstance(field, str):
-                raise TypeError(f"Custom field '{field}' is not a string")
-        return outdict, custom_fields
-
-    return wrapper
 
 REQUIRED_COLS = ["experiment", "dataset"]
 CSV_FILE_POSTFIX = "__perf"
+METRIC_PREFIX = "metric_"
 
-class PerfCalc(ABC): # Abstract base class for performance calculation
-
+class PerfCalc(ABC):  # Abstract base class for performance calculation
     @abstractmethod
-    def get_experiment_name(self):
+    def get_experiment_name(self) -> str:
         """
         Return the name of the experiment.
         This function should be overridden by the subclass if needed.
@@ -73,7 +29,7 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         pass
 
     @abstractmethod
-    def get_dataset_name(self):
+    def get_dataset_name(self) -> str:
         """
         Return the name of the dataset.
         This function should be overridden by the subclass if needed.
@@ -81,161 +37,135 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         pass
 
     @abstractmethod
-    def get_metrics_info(self):
+    def get_metric_backend(self) -> MetricsBackend:
         """
         Return a list of metric names to be used for performance calculation OR a dictionaray with keys as metric names and values as metric instances of torchmetrics.Metric. For example: {"accuracy": Accuracy(), "precision": Precision()}
 
         """
         pass
 
-    def calc_exp_outdict_custom_fields(self, outdict, *args, **kwargs):
-        """Can be overridden by the subclass to add custom fields to the output dictionary.
-        ! must return the modified outdict, and a ordered list of custom fields to be added to the output dictionary.
-        """
-        return outdict, []
-
-    # ! can be override, but ONLY if torchmetrics are used
-    # Prepare the exp data for torch metrics.
-    def prepare_torch_metrics_exp_data(self, metric_names, *args, **kwargs):
-        """
-        Prepare the data for metrics.
-        This function should be overridden by the subclass if needed.
-        Must return a dictionary with keys as metric names and values as the data to be used for those metrics.
-        NOTE: that the data (for each metric) must be in the format expected by the torchmetrics instance (for that metric). E.g: {"accuracy": {"preds": [...], "target": [...]}, ...} since torchmetrics expects the data in a specific format.
-        """
-        pass
-
-    def __validate_metrics_info(self, metrics_info):
-        """
-        Validate the metrics_info to ensure it is a list or a dictionary with valid metric names and instances.
-        """
-        if not isinstance(metrics_info, (list, dict)):
-            raise TypeError(f"Metrics info must be a list or a dictionary, got {type(metrics_info).__name__}")
-
-        if isinstance(metrics_info, dict):
-            for k, v in metrics_info.items():
-                if not isinstance(k, str):
-                    raise TypeError(f"Key '{k}' is not a string")
-                if not isinstance(v, Metric):
-                    raise TypeError(f"Value for key '{k}' is not a torchmetrics.Metric (got {type(v).__name__})")
-        elif isinstance(metrics_info, list):
-            for metric in metrics_info:
-                if not isinstance(metric, str):
-                    raise TypeError(f"Metric '{metric}' is not a string")
-        return metrics_info
-    def __calc_exp_perf_metrics(self, *args, **kwargs):
-        """
-        Calculate the performance metrics for the experiment.
-        """
-        metrics_info = self.__validate_metrics_info(self.get_metrics_info())
-        USED_TORCHMETRICS = isinstance(metrics_info, dict)
-        metric_names = metrics_info if isinstance(metrics_info, list) else list(metrics_info.keys())
-        out_dict = {metric: None for metric in metric_names}
-        out_dict["dataset"] = self.get_dataset_name()
-        out_dict["experiment"] = self.get_experiment_name()
-        out_dict, custom_fields = self.calc_exp_outdict_custom_fields(
-            outdict=out_dict, *args, **kwargs
-        )
-        if USED_TORCHMETRICS:
-            torch_metrics_dict = self.get_metrics_info()
-            all_metric_data = self.prepare_torch_metrics_exp_data(
-                metric_names, *args, **kwargs
+    def valid_proc_extra_data(
+        self, proc_extra_data
+    ):
+        # make sure that all items in proc_extra_data are dictionaries, with same keys
+        if proc_extra_data is None or len(proc_extra_data) == 0:
+            return
+        if not all(isinstance(item, dict) for item in proc_extra_data):
+            raise TypeError("All items in proc_extra_data must be dictionaries")
+
+        if not all(item.keys() == proc_extra_data[0].keys() for item in proc_extra_data):
+            raise ValueError("All dictionaries in proc_extra_data must have the same keys")
+
+    def valid_proc_metric_raw_data(
+        self, metric_names, proc_metric_raw_data
+    ):
+        # make sure that all items in proc_metric_raw_data are dictionaries, with same keys as metric_names
+        assert isinstance(proc_metric_raw_data, list) and len(proc_metric_raw_data) > 0, \
+            "raw_data_for_metrics must be a non-empty list of dictionaries"
+
+        # make sure that all items in proc_metric_raw_data are dictionaries with keys as metric_names
+        if not all(isinstance(item, dict) for item in proc_metric_raw_data):
+            raise TypeError("All items in raw_data_for_metrics must be dictionaries")
+        if not all( set(item.keys()) == set(metric_names) for item in proc_metric_raw_data):
+            raise ValueError(
+                "All dictionaries in raw_data_for_metrics must have the same keys as metric_names"
             )
-            metric_col_names = []
-            for metric in metric_names:
-                if metric not in all_metric_data:
-                    raise ValueError(f"Metric '{metric}' not found in provided data.")
-                tmetric = torch_metrics_dict[metric]  # torchmetrics instance
-                metric_data = all_metric_data[metric]  # should be a dict of args/kwargs
-                # Inspect expected parameters for the metric's update() method
-                sig = inspect.signature(tmetric.update)
-                expected_args = list(sig.parameters.values())
-                # Prepare args in correct order
-                if isinstance(metric_data, dict):
-                    # Match dict keys to parameter names
-                    args = [metric_data[param.name] for param in expected_args]
-                elif isinstance(metric_data, (list, tuple)):
-                    args = metric_data
-                else:
-                    raise TypeError(f"Unsupported data format for metric '{metric}'")
-
-                # Call update and compute
-                if len(expected_args) == 1:
-                    tmetric.update(args)  # pass as single argument
-                else:
-                    tmetric.update(*args)  # unpack multiple arguments
-                computed_value = tmetric.compute()
-                # ensure the computed value converted to a scala value or list array
-                if isinstance(computed_value, torch.Tensor):
-                    if computed_value.numel() == 1:
-                        computed_value = computed_value.item()
-                    else:
-                        computed_value = computed_value.tolist()
-                col_name = f"metric_{metric}" if "metric_" not in metric else metric
-                metric_col_names.append(col_name)
-                out_dict[col_name] = computed_value
-        else:
-            # If torchmetrics are not used, calculate metrics using the custom method
-            metric_rs_dict = self.calc_exp_perf_metrics(
-                metric_names, *args, **kwargs)
-            for metric in metric_names:
-                if metric not in metric_rs_dict:
-                    raise ValueError(f"Metric '{metric}' not found in provided data.")
-                col_name = f"metric_{metric}" if "metric_" not in metric else metric
-                out_dict[col_name] = metric_rs_dict[metric]
-            metric_col_names = [f"metric_{metric}" for metric in metric_names]
-        ordered_cols = REQUIRED_COLS + custom_fields + metric_col_names
-        # create a new ordered dictionary with the correct order
-        out_dict = OrderedDict((col, out_dict[col]) for col in ordered_cols if col in out_dict)
-        return out_dict
 
     # ! only need to override this method if torchmetrics are not used
-    def calc_exp_perf_metrics(self, metric_names, *args, **kwargs):
-        """
-        Calculate the performance metrics for the experiment, but not using torchmetrics.
-        This function should be overridden by the subclass if needed.
-        Must return a dictionary with keys as metric names and values as the calculated metrics.
-        """
-        raise NotImplementedError("calc_exp_perf_metrics() must be overridden by the subclass if torchmetrics are not used.")
+    def calc_exp_perf_metrics(
+        self, metric_names, raw_metrics_data, extra_data=None, *args, **kwargs
+    ):
+        assert isinstance(raw_metrics_data, dict) or isinstance(raw_metrics_data, list), \
+            "raw_data_for_metrics must be a dictionary or a list"
+
+        if extra_data is not None:
+            assert isinstance(extra_data, type(raw_metrics_data)), \
+                "extra_data must be of the same type as raw_data_for_metrics (dict or list)"
+        # prepare raw_metric data for processing
+        proc_metric_raw_data_ls = raw_metrics_data if isinstance(raw_metrics_data, list) else [raw_metrics_data.copy()]
+        self.valid_proc_metric_raw_data(metric_names, proc_metric_raw_data_ls)
+        # prepare extra data for processing
+        proc_extra_data_ls = []
+        if extra_data is not None:
+            proc_extra_data_ls = extra_data if isinstance(extra_data, list) else [extra_data.copy()]
+            assert len(proc_extra_data_ls) == len(proc_metric_raw_data_ls), \
+                "extra_data must have the same length as raw_data_for_metrics if it is a list"
+        # validate the extra_data
+        self.valid_proc_extra_data(proc_extra_data_ls)
+
+        # calculate the metrics output results
+        metrics_backend = self.get_metric_backend()
+        proc_outdict_list = []
+        for idx, raw_metrics_data in enumerate(proc_metric_raw_data_ls):
+            out_dict = {
+                "dataset": self.get_dataset_name(),
+                "experiment": self.get_experiment_name(),
+            }
+            custom_fields = []
+            if len(proc_extra_data_ls)> 0:
+                # add extra data to the output dictionary
+                extra_data_item = proc_extra_data_ls[idx]
+                out_dict.update(extra_data_item)
+                custom_fields = list(extra_data_item.keys())
+            metric_results = metrics_backend.calc_metrics(
+                metrics_data_dict=raw_metrics_data, *args, **kwargs
+            )
+            metric_results_prefix = {f"metric_{k}": v for k, v in metric_results.items()}
+            out_dict.update(metric_results_prefix)
+            ordered_cols = (
+                REQUIRED_COLS + custom_fields + list(metric_results_prefix.keys())
+            )
+            out_dict = OrderedDict(
+                (col, out_dict[col]) for col in ordered_cols if col in out_dict
+            )
+            proc_outdict_list.append(out_dict)
 
+        return proc_outdict_list
 
     #! custom kwargs:
     #! outfile - if provided, will save the output to a CSV file with the given path
     #! outdir - if provided, will save the output to a CSV file in the given directory with a generated filename
     #! return_df - if True, will return a DataFrame instead of a dictionary
-
-    def calc_save_exp_perfs(self, *args, **kwargs):
+    def calc_and_save_exp_perfs(
+        self,
+        raw_metrics_data: Union[List[dict], dict],
+        extra_data: Optional[Union[List[dict], dict]] = None,
+        *args,
+        **kwargs,
+    ) -> Tuple[Union[List[OrderedDict], pd.DataFrame], Optional[str]]:
         """
         Calculate the metrics.
         This function should be overridden by the subclass if needed.
         Must return a dictionary with keys as metric names and values as the calculated metrics.
         """
-        out_dict = self.__calc_exp_perf_metrics(*args, **kwargs)
-        # pprint(f"Output Dictionary: {out_dict}")
-        # check if any kwargs named "outfile"
+        metric_names = self.get_metric_backend().metric_names
+        out_dict_list = self.calc_exp_perf_metrics(
+            metric_names=metric_names, raw_metrics_data=raw_metrics_data,
+            extra_data=extra_data,
+            *args, **kwargs
+        )
         csv_outfile = kwargs.get("outfile", None)
         if csv_outfile is not None:
             filePathNoExt, _ = os.path.splitext(csv_outfile)
             # pprint(f"CSV Outfile Path (No Ext): {filePathNoExt}")
-            csv_outfile = f'{filePathNoExt}{CSV_FILE_POSTFIX}.csv'
+            csv_outfile = f"{filePathNoExt}{CSV_FILE_POSTFIX}.csv"
         elif "outdir" in kwargs:
             csvoutdir = kwargs["outdir"]
             csvfilename = f"{now_str()}_{self.get_dataset_name()}_{self.get_experiment_name()}_{CSV_FILE_POSTFIX}.csv"
             csv_outfile = os.path.join(csvoutdir, csvfilename)
 
         # convert out_dict to a DataFrame
-        df = pd.DataFrame([out_dict])
+        df = pd.DataFrame(out_dict_list)
         # get the orders of the columns as the orders or the keys in out_dict
-        ordered_cols = list(out_dict.keys())
+        ordered_cols = list(out_dict_list[0].keys())
         df = df[ordered_cols]  # reorder columns
-
         if csv_outfile:
             df.to_csv(csv_outfile, index=False, sep=";", encoding="utf-8")
         return_df = kwargs.get("return_df", False)
-        if return_df: # return DataFrame instead of dict if requested
+        if return_df:  # return DataFrame instead of dict if requested
             return df, csv_outfile
         else:
-            return out_dict, csv_outfile
+            return out_dict_list, csv_outfile
 
     @staticmethod
     def default_exp_csv_filter_fn(exp_file_name: str) -> bool:
@@ -247,29 +177,37 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
 
     @classmethod
     def gen_perf_report_for_multip_exps(
-        cls, indir: str, exp_csv_filter_fn=default_exp_csv_filter_fn, csv_sep=";"
+        cls, indir: str, exp_csv_filter_fn=default_exp_csv_filter_fn, include_file_name=False, csv_sep=";"
     ) -> PerfTB:
         """
         Generate a performance report by scanning experiment subdirectories.
         Must return a dictionary with keys as metric names and values as performance tables.
         """
-        def get_df_for_all_exp_perf(csv_perf_files, csv_sep=';'):
+        def get_df_for_all_exp_perf(csv_perf_files, csv_sep=";"):
             """
             Create a single DataFrame from all CSV files.
             Assumes all CSV files MAY have different metrics
             """
             cols = []
+            FILE_NAME_COL = "file_name" if include_file_name else None
+
             for csv_file in csv_perf_files:
                 temp_df = pd.read_csv(csv_file, sep=csv_sep)
+                if FILE_NAME_COL:
+                    temp_df[FILE_NAME_COL] = fs.get_file_name(csv_file, split_file_ext=False)
+                    # csvfile.fn_display_df(temp_df)
                 temp_df_cols = temp_df.columns.tolist()
                 for col in temp_df_cols:
                     if col not in cols:
                         cols.append(col)
+
             df = pd.DataFrame(columns=cols)
             for csv_file in csv_perf_files:
                 temp_df = pd.read_csv(csv_file, sep=csv_sep)
+                if FILE_NAME_COL:
+                    temp_df[FILE_NAME_COL] = fs.get_file_name(csv_file, split_file_ext=False)
                 # Drop all-NA columns to avoid dtype inconsistency
-                temp_df = temp_df.dropna(axis=1, how='all')
+                temp_df = temp_df.dropna(axis=1, how="all")
                 # ensure all columns are present in the final DataFrame
                 for col in cols:
                     if col not in temp_df.columns:
@@ -277,24 +215,36 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
                 df = pd.concat([df, temp_df], ignore_index=True)
             # assert that REQUIRED_COLS are present in the DataFrame
             # pprint(df.columns.tolist())
-            for col in REQUIRED_COLS:
+            sticky_cols = REQUIRED_COLS + ([FILE_NAME_COL] if include_file_name else []) # columns that must always be present
+            for col in sticky_cols:
                 if col not in df.columns:
-                    raise ValueError(f"Required column '{col}' is missing from the DataFrame. REQUIRED_COLS = {REQUIRED_COLS}")
-            metric_cols = [col for col in df.columns if col.startswith('metric_')]
-            assert len(metric_cols) > 0, "No metric columns found in the DataFrame. Ensure that the CSV files contain metric columns starting with 'metric_'."
-            final_cols = REQUIRED_COLS + metric_cols
+                    raise ValueError(
+                        f"Required column '{col}' is missing from the DataFrame. REQUIRED_COLS = {sticky_cols}"
+                    )
+            metric_cols = [col for col in df.columns if col.startswith(METRIC_PREFIX)]
+            assert (
+                len(metric_cols) > 0
+            ), "No metric columns found in the DataFrame. Ensure that the CSV files contain metric columns starting with 'metric_'."
+            final_cols = sticky_cols + metric_cols
             df = df[final_cols]
+            # !hahv debug
+            pprint("------ Final DataFrame Columns ------")
+            csvfile.fn_display_df(df)
             # ! validate all rows in df before returning
             # make sure all rows will have at least values for REQUIRED_COLS and at least one metric column
             for index, row in df.iterrows():
-                if not all(col in row and pd.notna(row[col]) for col in REQUIRED_COLS):
-                    raise ValueError(f"Row {index} is missing required columns or has NaN values in required columns: {row}")
+                if not all(col in row and pd.notna(row[col]) for col in sticky_cols):
+                    raise ValueError(
+                        f"Row {index} is missing required columns or has NaN values in required columns: {row}"
+                    )
                 if not any(pd.notna(row[col]) for col in metric_cols):
                     raise ValueError(f"Row {index} has no metric values: {row}")
             # make sure these is no (experiment, dataset) pair that is duplicated
-            duplicates = df.duplicated(subset=['experiment', 'dataset'], keep=False)
+            duplicates = df.duplicated(subset=sticky_cols, keep=False)
             if duplicates.any():
-                raise ValueError("Duplicate (experiment, dataset) pairs found in the DataFrame. Please ensure that each experiment-dataset combination is unique.")
+                raise ValueError(
+                    "Duplicate (experiment, dataset) pairs found in the DataFrame. Please ensure that each experiment-dataset combination is unique."
+                )
             return df
 
         def mk_perftb_report(df):
@@ -304,9 +254,9 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
             """
             perftb = PerfTB()
             # find all "dataset" values (unique)
-            dataset_names = list(df['dataset'].unique())
-            # find all columns that start with "metric_"
-            metric_cols = [col for col in df.columns if col.startswith('metric_')]
+            dataset_names = list(df["dataset"].unique())
+            # find all columns that start with METRIC_PREFIX
+            metric_cols = [col for col in df.columns if col.startswith(METRIC_PREFIX)]
 
             # Determine which metrics are associated with each dataset.
             # Since a dataset may appear in multiple rows and may not include all metrics in each, identify the row with the same dataset that contains the most non-NaN metric values. The set of metrics for that dataset is defined by the non-NaN metrics in that row.
@@ -316,7 +266,11 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
                 dataset_rows = df[df["dataset"] == dataset_name]
                 # Find the row with the most non-NaN metric values
                 max_non_nan_row = dataset_rows[metric_cols].count(axis=1).idxmax()
-                metrics_for_dataset = dataset_rows.loc[max_non_nan_row, metric_cols].dropna().index.tolist()
+                metrics_for_dataset = (
+                    dataset_rows.loc[max_non_nan_row, metric_cols]
+                    .dropna()
+                    .index.tolist()
+                )
                 dataset_metrics[dataset_name] = metrics_for_dataset
 
             for dataset_name, metrics in dataset_metrics.items():
@@ -324,11 +278,11 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
                 perftb.add_dataset(dataset_name, metrics)
 
             for _, row in df.iterrows():
-                dataset_name = row['dataset']
+                dataset_name = row["dataset"]
                 ds_metrics = dataset_metrics.get(dataset_name)
                 if dataset_name in dataset_metrics:
                     # Add the metrics for this row to the performance table
-                    exp_name = row.get('experiment')
+                    exp_name = row.get("experiment")
                     exp_metric_values = {}
                     for metric in ds_metrics:
                         if metric in row and pd.notna(row[metric]):
@@ -336,7 +290,7 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
                     perftb.add_experiment(
                         experiment_name=exp_name,
                         dataset_name=dataset_name,
-                        metrics=exp_metric_values
+                        metrics=exp_metric_values,
                     )
 
             return perftb
@@ -351,9 +305,7 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
             if os.path.isdir(os.path.join(indir, d))
         ]
         if len(exp_dirs) == 0:
-            csv_perf_files = glob.glob(
-                os.path.join(indir, f"*.csv")
-            )
+            csv_perf_files = glob.glob(os.path.join(indir, f"*.csv"))
             csv_perf_files = [
                 file_item
                 for file_item in csv_perf_files
@@ -364,13 +316,9 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
             # Collect all matching CSV files in those subdirs
             for exp_dir in exp_dirs:
                 # pprint(f"Searching in experiment directory: {exp_dir}")
-                matched = glob.glob(
-                    os.path.join(exp_dir, f"*.csv")
-                )
+                matched = glob.glob(os.path.join(exp_dir, f"*.csv"))
                 matched = [
-                    file_item
-                    for file_item in matched
-                    if exp_csv_filter_fn(file_item)
+                    file_item for file_item in matched if exp_csv_filter_fn(file_item)
                 ]
                 csv_perf_files.extend(matched)
 
@@ -378,9 +326,11 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
             len(csv_perf_files) > 0
         ), f"No CSV files matching pattern '{exp_csv_filter_fn}' found in the experiment directories."
 
-        assert len(csv_perf_files) > 0, f"No CSV files matching pattern '{exp_csv_filter_fn}' found in the experiment directories."
+        assert (
+            len(csv_perf_files) > 0
+        ), f"No CSV files matching pattern '{exp_csv_filter_fn}' found in the experiment directories."
 
         all_exp_perf_df = get_df_for_all_exp_perf(csv_perf_files, csv_sep=csv_sep)
-        csvfile.fn_display_df(all_exp_perf_df)
+        # csvfile.fn_display_df(all_exp_perf_df)
         perf_tb = mk_perftb_report(all_exp_perf_df)
-        return perf_tb
+        return perf_tb

{halib-0.1.65.dist-info → halib-0.1.67.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: halib
-Version: 0.1.65
+Version: 0.1.67
 Summary: Small library for common tasks
 Author: Hoang Van Ha
 Author-email: hoangvanhauit@gmail.com
@@ -52,7 +52,7 @@ Dynamic: summary
 
 Helper package for coding and automation
 
-**Version 0.1.65**
+**Version 0.1.67**
 
 + now use `uv` for venv management
 + `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation

{halib-0.1.65.dist-info → halib-0.1.67.dist-info}/RECORD

@@ -30,7 +30,8 @@ halib/online/projectmake.py,sha256=Zrs96WgXvO4nIrwxnCOletL4aTBge-EoF0r7hpKO1w8,4
 halib/research/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 halib/research/benchquery.py,sha256=FuKnbWQtCEoRRtJAfN-zaN-jPiO_EzsakmTOMiqi7GQ,4626
 halib/research/dataset.py,sha256=QU0Hr5QFb8_XlvnOMgC9QJGIpwXAZ9lDd0RdQi_QRec,6743
-halib/research/perfcalc.py,sha256=F1BYbxQohbS7u3iRtqnKgPmMrWneV6_bEdBumto8h58,18403
+halib/research/metrics.py,sha256=Xgv0GUGo-o-RJaBOmkRCRpQJaYijF_1xeKkyYU_Bv4U,5249
+halib/research/perfcalc.py,sha256=qDa0sqfpWrwGZVJtjuUVFK7JX6j8xyXP9OnnfYmdamg,15898
 halib/research/perftb.py,sha256=vazU-dYBJhfc4sK4zFgxOvzeXGi-5TyPHCt20ItiWhY,30463
 halib/research/plot.py,sha256=-pDUk4z3C_GnyJ5zWmf-mGMdT4gaipVJWzIgcpIPiRk,9448
 halib/research/torchloader.py,sha256=yqUjcSiME6H5W210363HyRUrOi3ISpUFAFkTr1w4DCw,6503
@@ -48,8 +49,8 @@ halib/utils/gpu_mon.py,sha256=vD41_ZnmPLKguuq9X44SB_vwd9JrblO4BDzHLXZhhFY,2233
 halib/utils/listop.py,sha256=Vpa8_2fI0wySpB2-8sfTBkyi_A4FhoFVVvFiuvW8N64,339
 halib/utils/tele_noti.py,sha256=-4WXZelCA4W9BroapkRyIdUu9cUVrcJJhegnMs_WpGU,5928
 halib/utils/video.py,sha256=ZqzNVPgc1RZr_T0OlHvZ6SzyBpL7O27LtB86JMbBuR0,3059
-halib-0.1.65.dist-info/licenses/LICENSE.txt,sha256=qZssdna4aETiR8znYsShUjidu-U4jUT9Q-EWNlZ9yBQ,1100
-halib-0.1.65.dist-info/METADATA,sha256=clqI54I9dybegyKBsPwaJM7cYBOCLYHdzXHEpFAKv_4,5541
-halib-0.1.65.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-halib-0.1.65.dist-info/top_level.txt,sha256=7AD6PLaQTreE0Fn44mdZsoHBe_Zdd7GUmjsWPyQ7I-k,6
-halib-0.1.65.dist-info/RECORD,,
+halib-0.1.67.dist-info/licenses/LICENSE.txt,sha256=qZssdna4aETiR8znYsShUjidu-U4jUT9Q-EWNlZ9yBQ,1100
+halib-0.1.67.dist-info/METADATA,sha256=Zk22ct5W95qBzGkz0tNepuAdfUwPJTbVO7Nb4L_hFTQ,5541
+halib-0.1.67.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+halib-0.1.67.dist-info/top_level.txt,sha256=7AD6PLaQTreE0Fn44mdZsoHBe_Zdd7GUmjsWPyQ7I-k,6
+halib-0.1.67.dist-info/RECORD,,