Configure HEROS testing and demo notebook to use new v0.8.2 of skrebate

.gitignore

@@ -173,6 +173,8 @@ cython_debug/
 # PyPI configuration file
 .pypirc
 
+# VS Code
+.vscode/
 #Project specific
 output/
 .DS_Store

evaluation/datasets/unit_testing/unit_EK/Multiplexer6_MultiSWRFDB_Scores.csv

@@ -0,0 +1,7 @@
+Feature,Score
+A_0,0.03698526666089879
+A_1,0.03698526666089878
+R_0,-0.0067603240148978025
+R_1,-0.006760324014897812
+R_2,-0.006760324014897798
+R_3,-0.006760324014897799

evaluation/datasets/unit_testing/unit_EK/Multiplexer6_NA_MultiSWRFDB_Scores.csv

@@ -0,0 +1,7 @@
+Feature,Score
+A_0,0.04750169656556999
+A_1,0.04136667893466966
+R_0,-0.012803708788285396
+R_1,-0.000140838501968226
+R_2,-0.03255689170033571
+R_3,-0.020210843651172736

evaluation/datasets/unit_testing/unit_EK/Multiplexer6_feature_type_mix_MultiSWRFDB_Scores.csv

@@ -0,0 +1,7 @@
+Feature,Score
+A_0,0.023352064355566495
+A_1,-0.006675774085889205
+R_0,-0.005446260273540202
+R_1,-0.001021047845508582
+R_2,0.02249503731040649
+R_3,0.026399857724587535

evaluation/datasets/unit_testing/unit_EK/Multiplexer6_feature_type_mix_NA_MultiSWRFDB_Scores.csv

@@ -0,0 +1,7 @@
+Feature,Score
+A_0,0.003406041029592358
+A_1,-0.00609654412659257
+R_0,-0.011544289442236113
+R_1,-0.007067428280550776
+R_2,0.007229959454005406
+R_3,0.036719539100810356

evaluation/datasets/unit_testing/unit_EK/Multiplexer6_multiclass_MultiSWRFDB_Scores.csv

@@ -0,0 +1,7 @@
+Feature,Score
+A_0,0.40813339250514147
+A_1,0.40813339250514147
+R_0,-0.0015949013505510805
+R_1,-0.0015949013505511637
+R_2,-0.0015949013505511499
+R_3,-0.0015949013505510527

pyproject.toml

@@ -31,7 +31,7 @@ dependencies = [
 "matplotlib",
 "seaborn",
 "scipy",
-"skrebate==0.7",
+"skrebate",
 "networkx"]
 
 [project.optional-dependencies]

tests/test_heros.py

@@ -1,25 +1,26 @@
-#import pytest
+# To test using pytest run `pytest --log-cli-level=DEBUG` from the root folder`
+
 import os
 import pandas as pd
 from src.skheros.heros import HEROS
 from sklearn.metrics import classification_report
-from skrebate import MultiSURF # Install using: pip install skrebate==0.7
-
-# To test using pytest run `pytest --log-cli-level=DEBUG` from the root folder`
+#from skrebate import MultiSURF 
+from skrebate import MultiSWRFDB
 
-def get_EK(X,y,data_name,feature_names):
+def get_EK(X,y,data_name,feature_names,cat_feat_indexes):
     """ Calculates or loads expert knowledge (EK) scores for the given unit testing dataset """
     # Further data preparation for expert knowldge generation (i.e. balanced subsampling of training instances for faster run times)
-    score_path_name = 'evaluation/datasets/unit_testing/unit_EK/'+str(data_name)+'_MultiSURF_Scores.csv' #No need to change
-    # Calculate or load feature importance estimates with MultiSURF ------------------------------------
+    score_path_name = 'evaluation/datasets/unit_testing/unit_EK/'+str(data_name)+'_MultiSWRFDB_Scores.csv' #No need to change
+    # Calculate or load feature importance estimates with MultiSWRFDB ------------------------------------
     if not os.path.isfile(score_path_name):
-        print("Generating MultiSURF Scores:")
-        clf = MultiSURF(n_jobs=1).fit(X, y)
+        print("Generating MultiSWRFDB Scores:")
+        #clf = MultiSURF(n_jobs=1).fit(X, y)
+        clf = MultiSWRFDB(n_jobs=1,categorical_features=cat_feat_indexes).fit(X, y)
         ek = clf.feature_importances_
         score_data = pd.DataFrame({'Feature':feature_names,'Score':ek})
         score_data.to_csv(score_path_name,index=False)
     else: #load previously trained scores
-        print("Loading MultiSURF Scores")
+        print("Loading MultiSWRFDB Scores")
         loaded_data = pd.read_csv(score_path_name)
         ek = loaded_data['Score'].tolist()
     return ek
@@ -36,7 +37,7 @@ def test_6mux():
     cat_feat_indexes = list(range(X.shape[1])) #all feature are categorical so provide indexes 0-5 in this list for 6-bit multiplexer dataset
     X = X.values
     y = df[outcome_label].values #outcome values
-    ek = get_EK(X,y,data_name,feature_names)
+    ek = get_EK(X,y,data_name,feature_names,cat_feat_indexes)
     print(ek)
     heros = HEROS(outcome_type='class',iterations=10000,pop_size=500,cross_prob=0.8,mut_prob=0.04,nu=1,beta=0.2,theta_sel=0.5,
                 fitness_function='pareto',subsumption='both',rsl=0,feat_track=None, model_iterations=40,
@@ -65,7 +66,8 @@ def test_na():
     cat_feat_indexes = list(range(X.shape[1])) #all feature are categorical so provide indexes 0-5 in this list for 6-bit multiplexer dataset
     X = X.values
     y = df[outcome_label].values #outcome values
-    ek = [0.0527170745920746, 0.0527170745920746, 0.00982931998557, 0.0098293199855699, 0.0098293199855699, 0.00982931998557] #temporary overide until Rebate fixed
+    ek = get_EK(X,y,data_name,feature_names,cat_feat_indexes)
+    #ek = [0.0527170745920746, 0.0527170745920746, 0.00982931998557, 0.0098293199855699, 0.0098293199855699, 0.00982931998557] #temporary overide until Rebate fixed
     #ek = get_EK(X,y,data_name,feature_names)
     print(ek)
     heros = HEROS(outcome_type='class',iterations=10000,pop_size=500,cross_prob=0.8,mut_prob=0.04,nu=1,beta=0.2,theta_sel=0.5,
@@ -92,10 +94,10 @@ def test_mixed_feature_types():
     outcome_label = 'outcome'
     X = df.drop(outcome_label, axis=1)
     feature_names = X.columns 
-    cat_feat_indexes = [0,1,2,3] #all feature are categorical so provide indexes 0-5 in this list for 6-bit multiplexer dataset
+    cat_feat_indexes = [0,1,2,3] # only first four features in this dataset can be treated as categorical
     X = X.values
     y = df[outcome_label].values #outcome values
-    ek = get_EK(X,y,data_name,feature_names)
+    ek = get_EK(X,y,data_name,feature_names,cat_feat_indexes)
     print(ek)
     heros = HEROS(outcome_type='class',iterations=20000,pop_size=500,cross_prob=0.8,mut_prob=0.04,nu=1,beta=0.2,theta_sel=0.5,
                 fitness_function='pareto',subsumption='both',rsl=0,feat_track=None, model_iterations=40,
@@ -116,15 +118,16 @@ def test_mixed_feature_types_na():
     print("------------------------------------------------------")
     print("Test: 6-bit MUX with mixed feature types and NAs - binary and quantitative features and binary outcome")
     # Load and prepare data
-    data_name = 'Multiplexer6_NA'
+    data_name = 'Multiplexer6_feature_type_mix_NA'
     df = pd.read_csv('evaluation/datasets/unit_testing/'+str(data_name)+'.csv')
     outcome_label = 'outcome'
     X = df.drop(outcome_label, axis=1)
     feature_names = X.columns 
-    cat_feat_indexes = [0,1,2,3] #all feature are categorical so provide indexes 0-5 in this list for 6-bit multiplexer dataset
+    cat_feat_indexes = [0,1,2,3] # only first four features in this dataset can be treated as categorical
     X = X.values
     y = df[outcome_label].values #outcome values
-    ek = [0.0527170745920746, 0.0527170745920746, 0.00982931998557, 0.0098293199855699, 0.0098293199855699, 0.00982931998557] #temporary overide until Rebate fixed
+    ek = get_EK(X,y,data_name,feature_names,cat_feat_indexes)
+    #ek = [0.0527170745920746, 0.0527170745920746, 0.00982931998557, 0.0098293199855699, 0.0098293199855699, 0.00982931998557] #temporary overide until Rebate fixed
     #ek = get_EK(X,y,data_name,feature_names)
     print(ek)
     heros = HEROS(outcome_type='class',iterations=20000,pop_size=500,cross_prob=0.8,mut_prob=0.04,nu=1,beta=0.2,theta_sel=0.5,
@@ -154,7 +157,7 @@ def test_multiclass():
     cat_feat_indexes = list(range(X.shape[1])) #all feature are categorical so provide indexes 0-5 in this list for 6-bit multiplexer dataset
     X = X.values
     y = df[outcome_label].values #outcome values
-    ek = get_EK(X,y,data_name,feature_names)
+    ek = get_EK(X,y,data_name,feature_names,cat_feat_indexes)
     print(ek)
     heros = HEROS(outcome_type='class',iterations=10000,pop_size=500,cross_prob=0.8,mut_prob=0.04,nu=1,beta=0.2,theta_sel=0.5,
                 fitness_function='pareto',subsumption='both',rsl=0,feat_track=None, model_iterations=40,