Flexibility improvements

.pre-commit-config.yaml

@@ -11,14 +11,14 @@ repos:
     -   id: check-yaml
     -   id: check-toml
     -   id: check-added-large-files
--   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.12.5
-    hooks:
-    -   id: ruff-check
-    -   id: ruff-format
 -   repo: https://github.com/nestauk/pre-commit-hooks
     rev: v1.2.0
     hooks:
     -   id: nbstripout-preserve-timestamp
     -   id: jupytext-enforce-pairing
     -   id: jupytext-smart-sync
+-   repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.12.5
+    hooks:
+    -   id: ruff-check
+    -   id: ruff-format

daindex/main.py

@@ -141,8 +141,8 @@ class DAIndex(object):
     `evaluate_all_groups_by_models` methods can be called to calculate the DAI for all groups relative to a
     single reference group.
 
-    The results can be accessed using the `get_group_ratios` and `get_group_figures` methods, or printed using
-    the `present_results` and `present_all_results` methods.
+    The results can be accessed using the `get_group_ratio` and `get_group_figure` methods (with a reference and
+    other group), or printed using the `present_results` and `present_all_results` methods.
 
     Args:
         cohort: The DataFrame containing the data for the cohort.
@@ -160,7 +160,7 @@ class DAIndex(object):
         weight_sum_steps: The number of bins for the weighted sum of k-step cutoffs.
         n_jobs: The number of jobs to run in parallel.
         model_name: The name of the model to use in the plots.
-        decision_boundary: The decision boundary for the DA curve.
+        decision_boundary: The decision boundary for the DA curve. Defaults to 0.5.
 
     Methods:
         setup_groups: Set up the groups for the DAI calculation.
@@ -174,10 +174,16 @@ class DAIndex(object):
             model objects.
         present_results: Print the DAI results for a pair of groups.
         present_all_results: Print the DAI results for all group pairs.
-        get_group_ratios: Get the DAI ratios for a pair of groups.
-        get_group_figures: Get the DAI figures for a pair of groups.
+        get_group_ratio: Get the DAI ratios for a pair of groups.
+        get_group_figure: Get the DAI figure for a pair of groups.
         get_all_ratios: Get the DAI ratios for all group pairs.
         get_all_figures: Get the DAI figures for all group pairs.
+        get_group_step_samples: Get the number of samples used for each step for a specific group.
+        get_group_sub_optimal_steps: Get information about sub-optimal steps for a specific group.
+        get_group_failed_steps: Get information about failed steps for a specific group.
+        get_all_groups_step_samples: Get step sample information for all evaluated groups.
+        get_all_groups_sub_optimal_steps: Get sub-optimal step information for all evaluated groups.
+        get_all_groups_failed_steps: Get failed step information for all evaluated groups.
     """
 
     def __init__(
@@ -221,6 +227,10 @@ def __init__(
         self.group_ratios = {}
         self.group_figures = {}
         self.issues = []
+        # New attributes to track step information
+        self.group_sub_optimal_steps = {}  # Stores sub-optimal steps for each group
+        self.group_failed_steps = {}  # Stores failed steps for each group
+        self.group_step_samples = {}  # Stores sample counts for each step for each group
 
     def setup_groups(self, groups: Group | list[Group], group_col: str = None) -> None:
         if not isinstance(groups, list):
@@ -485,6 +495,16 @@ def process_step(s: int) -> tuple[float, int, float, bool, bool]:
             delayed(process_step)(s)
             for s in tqdm(self.step_scores, desc=f"Calculating k-steps for '{group}' group", position=1, leave=False)
         )
+
+        # Store step information internally
+        sub_opt_steps = {s[0]: s[1] for s in ret if s[3]}
+        failed_steps = {s[0]: s[1] for s in ret if s[4]}
+        step_samples = {s[0]: s[1] for s in ret}
+
+        self.group_sub_optimal_steps[group] = sub_opt_steps
+        self.group_failed_steps[group] = failed_steps
+        self.group_step_samples[group] = step_samples
+
         sub_opt_list = ", ".join([f"{s[0]}: {s[1]}" for s in ret if s[3]])
         failed_list = ", ".join([f"{s[0]}: {s[1]}" for s in ret if s[4]])
         if sub_opt_list or failed_list:
@@ -745,10 +765,10 @@ def present_all_results(self) -> None:
         for group_pair in self.group_figures.keys():
             self.present_results(*group_pair)
 
-    def get_group_ratios(self, reference_group: str, other_group: str) -> dict[str, float | str]:
+    def get_group_ratio(self, reference_group: str, other_group: str) -> dict[str, float | str]:
         return self.group_ratios[(reference_group, other_group)]
 
-    def get_group_figures(self, reference_group: str, other_group: str) -> plt.Figure:
+    def get_group_figure(self, reference_group: str, other_group: str) -> plt.Figure:
         return self.group_figures[(reference_group, other_group)]
 
     def get_all_ratios(self) -> pd.DataFrame:
@@ -758,3 +778,81 @@ def get_all_ratios(self) -> pd.DataFrame:
 
     def get_all_figures(self) -> dict[tuple[str, str], plt.Figure]:
         return self.group_figures
+
+    def get_group_step_samples(self, group: str) -> dict[float, int]:
+        """
+        Returns the number of samples used for each step for the specified group.
+
+        Args:
+            group: The name of the group to get step sample information for.
+
+        Returns:
+            A dictionary mapping step scores to the number of samples used.
+        """
+        if group not in self.group_step_samples:
+            raise ValueError(f"No step information available for group '{group}'. Run evaluation first.")
+        return self.group_step_samples[group]
+
+    def get_group_sub_optimal_steps(self, group: str) -> dict[float, int]:
+        """
+        Returns information about sub-optimal steps for the specified group.
+        These are steps where the number of samples was below the preferred threshold
+        but still acceptable.
+
+        Args:
+            group: The name of the group to get sub-optimal step information for.
+
+        Returns:
+            A dictionary mapping step scores to the number of samples used for
+            sub-optimal steps.
+        """
+        if group not in self.group_sub_optimal_steps:
+            raise ValueError(f"No step information available for group '{group}'. Run evaluation first.")
+        return self.group_sub_optimal_steps[group]
+
+    def get_group_failed_steps(self, group: str) -> dict[float, int]:
+        """
+        Returns information about failed steps for the specified group.
+        These are steps where the number of samples was below the minimum threshold
+        and the step was excluded from the analysis.
+
+        Args:
+            group: The name of the group to get failed step information for.
+
+        Returns:
+            A dictionary mapping step scores to the number of samples used for
+            failed steps.
+        """
+        if group not in self.group_failed_steps:
+            raise ValueError(f"No step information available for group '{group}'. Run evaluation first.")
+        return self.group_failed_steps[group]
+
+    def get_all_groups_step_samples(self) -> dict[str, dict[float, int]]:
+        """
+        Returns the number of samples used for each step for all evaluated groups.
+
+        Returns:
+            A dictionary mapping group names to dictionaries of step scores and
+            their corresponding sample counts.
+        """
+        return self.group_step_samples.copy()
+
+    def get_all_groups_sub_optimal_steps(self) -> dict[str, dict[float, int]]:
+        """
+        Returns information about sub-optimal steps for all evaluated groups.
+
+        Returns:
+            A dictionary mapping group names to dictionaries of step scores and
+            their corresponding sample counts for sub-optimal steps.
+        """
+        return self.group_sub_optimal_steps.copy()
+
+    def get_all_groups_failed_steps(self) -> dict[str, dict[float, int]]:
+        """
+        Returns information about failed steps for all evaluated groups.
+
+        Returns:
+            A dictionary mapping group names to dictionaries of step scores and
+            their corresponding sample counts for failed steps.
+        """
+        return self.group_failed_steps.copy()

daindex/util.py

@@ -86,8 +86,13 @@ def compare_two_groups(
 
 
 def area_under_curve(w_data: np.ndarray, decision_boundary: float = 0.5) -> tuple[float, float]:
-    """
-    calculate the area under curve - do NOT do interpolation
+    """Calculate the area under curve - do NOT do interpolation
+
+    Args:
+        w_data: The weighted data, should be a 2D array with shape (n, 3) where
+            the first column is the x values, the second column is the y values,
+            and the third column is the weights.
+        decision_boundary: The decision boundary for calculating the area under curve.
     """
     prev = None
     area = 0.0

pyproject.toml

@@ -11,7 +11,7 @@ authors = [
     { name = "Harrison Wilde", email = "h.wilde@ucl.ac.uk" },
     { name = "Honghan Wu", email = "honghan.wu@gmail.com" },
 ]
-version = "0.7.3"
+version = "0.8.0"
 description = "Deterioration Allocation Index Framework"
 license = "MIT"
 readme = "README.md"

tutorials/meps_tutorial.ipynb

@@ -494,9 +494,13 @@
     "    cohort=df,\n",
     "    groups=groups,\n",
     "    det_feature=det_feature,\n",
+    "    decision_boundary=0.75,  # Set decision boundary for DA curve\n",
     ")\n",
     "index.evaluate_all_groups_by_models(\n",
-    "    models=lr_orig_panel19, feature_list=feature_list, reference_group=\"White\", n_jobs=6\n",
+    "    models=lr_orig_panel19,\n",
+    "    feature_list=feature_list,\n",
+    "    reference_group=\"White\",\n",
+    "    n_jobs=1,  # Up n_jobs to parallelize if needed\n",
     ")"
    ]
   },
@@ -518,6 +522,24 @@
     "index.get_all_ratios()"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "index.get_group_ratio(\"White\", \"Asian\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "index.get_all_groups_failed_steps()"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -527,8 +549,11 @@
   }
  ],
  "metadata": {
+  "jupytext": {
+   "formats": "ipynb,py:percent"
+  },
   "kernelspec": {
-   "display_name": "daindex",
+   "display_name": ".venv",
    "language": "python",
    "name": "python3"
   },
@@ -542,7 +567,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.5"
+   "version": "3.10.18"
   }
  },
  "nbformat": 4,