sdufseval 1.0.2__tar.gz → 1.0.3__tar.gz

{sdufseval-1.0.2 → sdufseval-1.0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sdufseval
-Version: 1.0.2
+Version: 1.0.3
 Summary: Evaluation and Benchmark Tool for Feature Selection
 Project-URL: Homepage, https://github.com/mrajabinasab/FSEVAL
 Project-URL: Bug Tracker, https://github.com/mrajabinasab/FSEVAL/issues
@@ -118,7 +118,9 @@ Initializes the evalutation and benchmark object.
 | :--- | :--- | :--- |
 | **`output_dir`** | results | Folder where CSV result files are saved. |
 | **`cv`** | 5 | Cross-validation folds (supervised only). |
-| **`avg_steps`** | 10 | Number of random restarts / seeds to average over.|
+| **`avg_steps`** | 10 | Number of repetitions for stochastic methods.|
+| **`supervised_iter`** | 5 | Number of classifier's runs with different random seeds.|
+| **`unsupervised_iter`** | 10 | Number of clustering runs with different random seeds.|
 | **`eval_type`** | both | "supervised", "unsupervised", or "both". |
 | **`metrics`** | ["CLSACC", "NMI", "ACC", "AUC"] | Evaluation metrics to calculate. |
 | **`experiments`** | ["10Percent", "100Percent"] | Which feature ratio grids to evaluate. |

{sdufseval-1.0.2 → sdufseval-1.0.3}/README.md

@@ -95,7 +95,9 @@ Initializes the evalutation and benchmark object.
 | :--- | :--- | :--- |
 | **`output_dir`** | results | Folder where CSV result files are saved. |
 | **`cv`** | 5 | Cross-validation folds (supervised only). |
-| **`avg_steps`** | 10 | Number of random restarts / seeds to average over.|
+| **`avg_steps`** | 10 | Number of repetitions for stochastic methods.|
+| **`supervised_iter`** | 5 | Number of classifier's runs with different random seeds.|
+| **`unsupervised_iter`** | 10 | Number of clustering runs with different random seeds.|
 | **`eval_type`** | both | "supervised", "unsupervised", or "both". |
 | **`metrics`** | ["CLSACC", "NMI", "ACC", "AUC"] | Evaluation metrics to calculate. |
 | **`experiments`** | ["10Percent", "100Percent"] | Which feature ratio grids to evaluate. |

{sdufseval-1.0.2 → sdufseval-1.0.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sdufseval"
-version = "1.0.2"
+version = "1.0.3"
 authors = [
   { name="Muhammad Rajabinasab", email="muhammad.rajabinasab@outlook.com" },
 ]

{sdufseval-1.0.2 → sdufseval-1.0.3}/sdufseval/fseval.py

@@ -12,7 +12,9 @@ class FSEVAL:
     def __init__(self, 
                  output_dir="results", 
                  cv=5, 
-                 avg_steps=10, 
+                 avg_steps=10,
+                 supervised_iter=5,
+                 unsupervised_iter=10, 
                  eval_type="both", 
                  metrics=None, 
                  experiments=None):
@@ -22,6 +24,8 @@ class FSEVAL:
         self.output_dir = output_dir
         self.cv = cv
         self.avg_steps = avg_steps
+        self.supervised_iter = supervised_iter
+        self.unsupervised_iter = unsupervised_iter
         self.eval_type = eval_type
         
         # Metric configuration
@@ -68,7 +72,7 @@ class FSEVAL:
                 name = m_info['name']
                 fs_func = m_info['func']
                 # Stochastic methods run 10 times and average
-                repeats = 10 if m_info.get('stochastic', False) else 1
+                repeats = self.avg_steps if m_info.get('stochastic', False) else 1
                 
                 # Internal storage for current dataset results
                 ds_results = {s: {met: [] for met in self.selected_metrics} for s in self.scales}
@@ -91,11 +95,11 @@ class FSEVAL:
                             c_acc, nmi, acc, auc = np.nan, np.nan, np.nan, np.nan
                             
                             if self.eval_type in ["unsupervised", "both"]:
-                                c_acc, nmi = unsupervised_eval(X_subset, y, avg_steps=self.avg_steps)
+                                c_acc, nmi = unsupervised_eval(X_subset, y, avg_steps=self.unsupervised_iter)
                             
                             if self.eval_type in ["supervised", "both"]:
                                 # Passes classifier (None or instance) to eval.py
-                                acc, auc = supervised_eval(X_subset, y, classifier=classifier, cv=self.cv, avg_steps=self.avg_steps)
+                                acc, auc = supervised_eval(X_subset, y, classifier=classifier, cv=self.cv, avg_steps=self.supervised_iter)
 
                             # Map metrics to columns
                             mapping = {"CLSACC": c_acc, "NMI": nmi, "ACC": acc, "AUC": auc}