Deliverable files added

requirements.txt

@@ -1,4 +1,4 @@
 numpy
 pandas
 matplotlib
-
+scikit-learn

src/figures/tuning_results.csv

@@ -0,0 +1,2 @@
+config,preprocess,model,params,mean_acc,std_acc
+scale + svm,scale,svm,"{'kernel': 'rbf', 'C': 4, 'gamma': 'scale', 'class_weight': None}",0.8581719138625145,0.013348223889216927

src/output.txt

@@ -0,0 +1,185 @@
+/home/hoo2/Work/AUTh/PatternRecognition/Assignment_2025-26/.venv/bin/python /home/hoo2/Work/AUTh/PatternRecognition/Assignment_2025-26/src/partD.py all 
+[       scale] [       gnb] val_acc=0.7095
+[       scale] [        rf] val_acc=0.8205
+[       scale] [    logreg] val_acc=0.7730
+[       scale] [linear_svm] val_acc=0.7707
+[       scale] [       svm] val_acc=0.8593
+[       scale] [       mlp] val_acc=0.8382
+[       scale] [       knn] val_acc=0.8342
+[       scale] [  adaboost] val_acc=0.6832
+[scale_pca_66] [       gnb] val_acc=0.7524
+[scale_pca_66] [        rf] val_acc=0.8096
+[scale_pca_66] [    logreg] val_acc=0.7862
+[scale_pca_66] [linear_svm] val_acc=0.7736
+[scale_pca_66] [       svm] val_acc=0.8582
+[scale_pca_66] [       mlp] val_acc=0.8359
+[scale_pca_66] [       knn] val_acc=0.8370
+[scale_pca_66] [  adaboost] val_acc=0.6878
+[scale_pca_75] [       gnb] val_acc=0.7547
+[scale_pca_75] [        rf] val_acc=0.8130
+[scale_pca_75] [    logreg] val_acc=0.7839
+[scale_pca_75] [linear_svm] val_acc=0.7696
+[scale_pca_75] [       svm] val_acc=0.8565
+[scale_pca_75] [       mlp] val_acc=0.8216
+[scale_pca_75] [       knn] val_acc=0.8370
+[scale_pca_75] [  adaboost] val_acc=0.6878
+[scale_pca_85] [       gnb] val_acc=0.7501
+[scale_pca_85] [        rf] val_acc=0.8033
+[scale_pca_85] [    logreg] val_acc=0.7810
+[scale_pca_85] [linear_svm] val_acc=0.7662
+[scale_pca_85] [       svm] val_acc=0.8588
+[scale_pca_85] [       mlp] val_acc=0.8188
+[scale_pca_85] [       knn] val_acc=0.8388
+[scale_pca_85] [  adaboost] val_acc=0.6998
+
+=== Investigation summary ===
+model
+svm           0.859348
+knn           0.838765
+mlp           0.838193
+rf            0.820469
+logreg        0.786164
+linear_svm    0.773585
+gnb           0.754717
+adaboost      0.699828
+
+Selected top-3 models for further analysis: ['svm', 'knn', 'mlp']
+
+Best configuration overall: preprocess=scale, model=svm, val_acc=0.8593
+
+Classification report (best config):
+              precision    recall  f1-score   support
+
+           1       0.94      0.96      0.95       354
+           2       0.76      0.73      0.75       344
+           3       0.92      0.93      0.93       351
+           4       0.91      0.91      0.91       343
+           5       0.75      0.77      0.76       357
+
+    accuracy                           0.86      1749
+   macro avg       0.86      0.86      0.86      1749
+weighted avg       0.86      0.86      0.86      1749
+
+
+[TUNING] scale + rf (cv=5) ...
+[scale | rf] combo   1/1 mean=0.8228 params={'n_estimators': 400, 'max_depth': None, 'max_features': 'sqrt', 'min_samples_split': 4, 'min_samples_leaf': 1}
+  best mean_acc=0.8228 (std=0.0121) params={'n_estimators': 400, 'max_depth': None, 'max_features': 'sqrt', 'min_samples_split': 4, 'min_samples_leaf': 1}
+
+[TUNING] scale + mlp (cv=5) ...
+[scale | mlp] combo   1/1 mean=0.8407 params={'hidden_layer_sizes': (128,), 'alpha': 0.001, 'learning_rate_init': 0.01, 'activation': 'relu', 'solver': 'adam'}
+  best mean_acc=0.8407 (std=0.0098) params={'hidden_layer_sizes': (128,), 'alpha': 0.001, 'learning_rate_init': 0.01, 'activation': 'relu', 'solver': 'adam'}
+
+[TUNING] scale_pca_85 + knn (cv=5) ...
+[scale_pca_85 | knn] combo   1/1 mean=0.8313 params={'n_neighbors': 9, 'weights': 'distance', 'p': 2}
+  best mean_acc=0.8313 (std=0.0117) params={'n_neighbors': 9, 'weights': 'distance', 'p': 2}
+
+[TUNING] scale + svm (cv=5) ...
+[scale | svm] combo   1/1 mean=0.8582 params={'kernel': 'rbf', 'C': 4, 'gamma': 'scale', 'class_weight': None}
+  best mean_acc=0.8582 (std=0.0133) params={'kernel': 'rbf', 'C': 4, 'gamma': 'scale', 'class_weight': None}
+
+=== Tuning summary (best overall) ===
+{'name': 'scale + svm', 'preprocess_spec': {'type': 'pipeline', 'steps': [{'type': 'scaler', 'params': {}}]}, 'preprocess_name': 'scale', 'model': 'svm', 'params': {'kernel': 'rbf', 'C': 4, 'gamma': 'scale', 'class_weight': None}, 'mean_acc': 0.8581719138625145, 'std_acc': 0.013348223889216927}
+
+============================================================
+[FINAL - VALIDATION] scale + rf
+Confusion matrix:
+[[338   6   5   3   2]
+ [  4 239  11  12  78]
+ [ 11   2 316  21   1]
+ [  3  12  17 299  12]
+ [ 13  71   4   9 260]]
+
+Classification report:
+              precision    recall  f1-score   support
+
+           1       0.92      0.95      0.93       354
+           2       0.72      0.69      0.71       344
+           3       0.90      0.90      0.90       351
+           4       0.87      0.87      0.87       343
+           5       0.74      0.73      0.73       357
+
+    accuracy                           0.83      1749
+   macro avg       0.83      0.83      0.83      1749
+weighted avg       0.83      0.83      0.83      1749
+
+============================================================
+[FINAL] scale_rf: saved labelsX_scale_rf.npy shape=(6955,)
+
+============================================================
+[FINAL - VALIDATION] scale + mlp
+Confusion matrix:
+[[338   1   9   2   4]
+ [  5 244  13   7  75]
+ [ 10   3 320  16   2]
+ [  0  14  16 302  11]
+ [  8  74   1  16 258]]
+
+Classification report:
+              precision    recall  f1-score   support
+
+           1       0.94      0.95      0.95       354
+           2       0.73      0.71      0.72       344
+           3       0.89      0.91      0.90       351
+           4       0.88      0.88      0.88       343
+           5       0.74      0.72      0.73       357
+
+    accuracy                           0.84      1749
+   macro avg       0.83      0.84      0.83      1749
+weighted avg       0.83      0.84      0.84      1749
+
+============================================================
+[FINAL] scale_mlp: saved labelsX_scale_mlp.npy shape=(6955,)
+
+============================================================
+[FINAL - VALIDATION] scale_pca_85 + knn
+Confusion matrix:
+[[346   2   5   0   1]
+ [  5 193   9   7 130]
+ [ 19   1 319  11   1]
+ [  4   9  17 301  12]
+ [  8  33   1   6 309]]
+
+Classification report:
+              precision    recall  f1-score   support
+
+           1       0.91      0.98      0.94       354
+           2       0.81      0.56      0.66       344
+           3       0.91      0.91      0.91       351
+           4       0.93      0.88      0.90       343
+           5       0.68      0.87      0.76       357
+
+    accuracy                           0.84      1749
+   macro avg       0.85      0.84      0.84      1749
+weighted avg       0.85      0.84      0.84      1749
+
+============================================================
+[FINAL] scale_pca_85_knn: saved labelsX_scale_pca_85_knn.npy shape=(6955,)
+
+============================================================
+[FINAL - VALIDATION] scale + svm
+Confusion matrix:
+[[340   2   8   1   3]
+ [  3 251   9   6  75]
+ [  7   1 327  14   2]
+ [  0  12   9 311  11]
+ [ 11  63   1   8 274]]
+
+Classification report:
+              precision    recall  f1-score   support
+
+           1       0.94      0.96      0.95       354
+           2       0.76      0.73      0.75       344
+           3       0.92      0.93      0.93       351
+           4       0.91      0.91      0.91       343
+           5       0.75      0.77      0.76       357
+
+    accuracy                           0.86      1749
+   macro avg       0.86      0.86      0.86      1749
+weighted avg       0.86      0.86      0.86      1749
+
+============================================================
+[FINAL] scale_svm: saved labelsX_scale_svm.npy shape=(6955,)
+Saved labels to labelsX.npy with shape (6955,)
+
+Process finished with exit code 0
+

src/partA.py

@@ -197,6 +197,17 @@ def plot_gaussians_3d(
     ax.set_zlabel("pdf")
     plt.show()
 
+    # plt.figure(figsize=(6, 5))
+    # plt.scatter(X[:, 0], X[:, 1], s=10, alpha=0.35)
+    # plt.contour(Xgrid, Ygrid, Z, levels=8, linewidths=1.5)
+    #
+    # plt.title("Estimated Gaussian density (ML)")
+    # plt.xlabel("x₁")
+    # plt.ylabel("x₂")
+    #
+    # plt.tight_layout()
+    # plt.show()
+
 
 
 # --------------------------------------------------

src/partB.py

@@ -302,7 +302,7 @@ def plot_histogram_with_pdf(
     plt.plot(x_plot, pdf_true, label=f"True N({mu_true}, {var_true}) pdf")
     plt.xlabel("x")
     plt.ylabel("Density")
-    plt.title("Dataset2 histogram vs true N({mu_true}, {var_true}) pdf")
+    plt.title(f"Dataset2 histogram vs true N({mu_true}, {var_true}) pdf")
     plt.legend()
     plt.grid(True)
     plt.show()

src/toolbox.py

@@ -21,7 +21,8 @@ dataset1 = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/datase
 dataset2 = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/dataset2.csv")
 dataset3 = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/dataset3.csv")
 testset  = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/testset.csv")
-
+datasetTV   = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/datasetTV.csv")
+datasetTest = github_raw("hoo2", "PR-Assignment2025_26", "master", "datasets/datasetTest.csv")
 
 def load_csv(path, header=None):
     """