File size: 26,390 Bytes
08f115c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
"""
Lab 5: Data Mining Algorithms
==============================
BIM5021 - Nhà kho dữ liệu và Tích hợp
Chương 5: Thuật toán khai thác dữ liệu cốt lõi

Mục tiêu:
- Implement Apriori từ đầu (from scratch)
- Association Rules mining trên Olist order items
- Decision Tree (ID3-style) với scikit-learn
- K-Means Customer Segmentation (RFM)
- DBSCAN Anomaly Detection
- Naive Bayes cho review prediction

Yêu cầu: pip install pandas numpy scikit-learn matplotlib mlxtend
"""

import pandas as pd
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from collections import defaultdict
from itertools import combinations
import warnings
warnings.filterwarnings('ignore')


# ==============================================================================
# PHẦN 1: APRIORI ALGORITHM (From Scratch!)
# ==============================================================================

class AprioriFromScratch:
    """
    Implementation Apriori algorithm từ đầu.
    Tham khảo: Agrawal & Srikant, 1994
    """
    
    def __init__(self, min_support: float = 0.01, min_confidence: float = 0.5):
        self.min_support = min_support
        self.min_confidence = min_confidence
        self.frequent_itemsets = {}
        self.rules = []
    
    def _get_support(self, itemset: frozenset, transactions: list) -> float:
        """Tính support của 1 itemset."""
        count = sum(1 for t in transactions if itemset.issubset(t))
        return count / len(transactions)
    
    def _get_frequent_1_itemsets(self, transactions: list) -> dict:
        """Tìm tất cả frequent 1-itemsets."""
        item_counts = defaultdict(int)
        for t in transactions:
            for item in t:
                item_counts[frozenset([item])] += 1
        
        n = len(transactions)
        return {
            itemset: count / n
            for itemset, count in item_counts.items()
            if count / n >= self.min_support
        }
    
    def _generate_candidates(self, prev_frequent: dict, k: int) -> set:
        """Generate candidate k-itemsets từ (k-1)-itemsets."""
        items = list(prev_frequent.keys())
        candidates = set()
        
        for i in range(len(items)):
            for j in range(i + 1, len(items)):
                # Union of two (k-1)-itemsets
                candidate = items[i] | items[j]
                if len(candidate) == k:
                    # Apriori pruning: check all (k-1) subsets are frequent
                    all_subsets_frequent = True
                    for item in candidate:
                        subset = candidate - frozenset([item])
                        if subset not in prev_frequent:
                            all_subsets_frequent = False
                            break
                    
                    if all_subsets_frequent:
                        candidates.add(candidate)
        
        return candidates
    
    def fit(self, transactions: list):
        """Chạy thuật toán Apriori."""
        print(f"\n  [APRIORI] min_support={self.min_support}, "
              f"min_confidence={self.min_confidence}")
        print(f"  [APRIORI] {len(transactions)} transactions")
        
        # Step 1: L1
        L1 = self._get_frequent_1_itemsets(transactions)
        self.frequent_itemsets[1] = L1
        print(f"  [L1] {len(L1)} frequent 1-itemsets")
        
        # Step 2: Iterate k = 2, 3, ...
        k = 2
        prev_frequent = L1
        
        while prev_frequent:
            # Generate candidates
            candidates = self._generate_candidates(prev_frequent, k)
            
            if not candidates:
                break
            
            # Count support for candidates
            Lk = {}
            for candidate in candidates:
                support = self._get_support(candidate, transactions)
                if support >= self.min_support:
                    Lk[candidate] = support
            
            if Lk:
                self.frequent_itemsets[k] = Lk
                print(f"  [L{k}] {len(Lk)} frequent {k}-itemsets "
                      f"(from {len(candidates)} candidates)")
            
            prev_frequent = Lk
            k += 1
        
        # Generate rules
        self._generate_rules(transactions)
        
        return self
    
    def _generate_rules(self, transactions: list):
        """Generate association rules từ frequent itemsets."""
        self.rules = []
        
        for k in range(2, max(self.frequent_itemsets.keys()) + 1):
            if k not in self.frequent_itemsets:
                continue
                
            for itemset, support in self.frequent_itemsets[k].items():
                # Generate all non-empty proper subsets
                items = list(itemset)
                for i in range(1, len(items)):
                    for antecedent_items in combinations(items, i):
                        antecedent = frozenset(antecedent_items)
                        consequent = itemset - antecedent
                        
                        # Find antecedent support
                        ant_support = None
                        ant_k = len(antecedent)
                        if ant_k in self.frequent_itemsets:
                            ant_support = self.frequent_itemsets[ant_k].get(antecedent)
                        
                        if ant_support is None or ant_support == 0:
                            continue
                        
                        confidence = support / ant_support
                        
                        if confidence >= self.min_confidence:
                            # Calculate lift
                            cons_k = len(consequent)
                            cons_support = self.frequent_itemsets.get(cons_k, {}).get(
                                consequent, 0
                            )
                            lift = confidence / cons_support if cons_support > 0 else 0
                            
                            self.rules.append({
                                'antecedent': antecedent,
                                'consequent': consequent,
                                'support': round(support, 4),
                                'confidence': round(confidence, 4),
                                'lift': round(lift, 4),
                            })
        
        # Sort by lift
        self.rules.sort(key=lambda x: x['lift'], reverse=True)
        print(f"  [RULES] {len(self.rules)} rules generated "
              f"(confidence >= {self.min_confidence})")
    
    def print_rules(self, top_n: int = 10):
        """In top rules."""
        print(f"\n  Top {min(top_n, len(self.rules))} Association Rules:")
        print(f"  {'Antecedent':<30} {'Consequent':<20} "
              f"{'Support':>8} {'Confidence':>10} {'Lift':>8}")
        print(f"  {'-'*80}")
        
        for rule in self.rules[:top_n]:
            ant = ', '.join(sorted(rule['antecedent']))
            cons = ', '.join(sorted(rule['consequent']))
            print(f"  {ant:<30}{cons:<16} "
                  f"{rule['support']:>8.4f} {rule['confidence']:>10.4f} "
                  f"{rule['lift']:>8.2f}")


def demo_apriori():
    """Demo Apriori trên simulated Olist data."""
    
    print("\n" + "=" * 70)
    print("  PART 1: APRIORI - ASSOCIATION RULES MINING")
    print("=" * 70)
    
    np.random.seed(42)
    
    # Simulate market basket data (product categories from Olist)
    categories = [
        'bed_bath_table', 'health_beauty', 'sports_leisure',
        'furniture_decor', 'computers_accessories', 'housewares',
        'watches_gifts', 'telephony', 'garden_tools', 'auto',
        'cool_stuff', 'perfumery', 'toys', 'baby', 'electronics'
    ]
    
    # Generate transactions with realistic patterns
    transactions = []
    for _ in range(500):
        n_items = np.random.choice([1, 2, 3, 4], p=[0.4, 0.35, 0.2, 0.05])
        basket = set()
        
        # Add correlated items
        r = np.random.random()
        if r < 0.3:
            basket.update(['bed_bath_table', 'housewares'])
        elif r < 0.5:
            basket.update(['health_beauty', 'perfumery'])
        elif r < 0.6:
            basket.update(['computers_accessories', 'telephony'])
        elif r < 0.7:
            basket.update(['baby', 'toys'])
        
        # Fill remaining with random
        while len(basket) < n_items:
            basket.add(np.random.choice(categories))
        
        transactions.append(frozenset(basket))
    
    # Run Apriori
    apriori = AprioriFromScratch(min_support=0.03, min_confidence=0.3)
    apriori.fit(transactions)
    apriori.print_rules(top_n=15)
    
    # Summary
    print(f"\n  Summary:")
    for k, items in apriori.frequent_itemsets.items():
        print(f"    L{k}: {len(items)} frequent {k}-itemsets")


# ==============================================================================
# PHẦN 2: DECISION TREE (scikit-learn + visualization)
# ==============================================================================

def demo_decision_tree():
    """Decision Tree cho dự đoán customer satisfaction."""
    
    from sklearn.tree import DecisionTreeClassifier, export_text
    from sklearn.model_selection import train_test_split
    from sklearn.metrics import classification_report, accuracy_score
    
    print("\n" + "=" * 70)
    print("  PART 2: DECISION TREE - CUSTOMER SATISFACTION")
    print("=" * 70)
    
    np.random.seed(42)
    n = 1000
    
    # Create dataset
    delivery_days = np.random.exponential(10, n)
    price = np.random.lognormal(4, 1, n)
    freight_ratio = np.random.uniform(0.05, 0.5, n)
    weight_kg = np.random.lognormal(1, 0.8, n)
    is_weekend = np.random.binomial(1, 0.3, n)
    
    # Target: satisfied (review >= 4) based on features
    satisfaction_prob = (
        0.8 
        - 0.02 * np.clip(delivery_days, 0, 30)
        - 0.001 * np.clip(price, 0, 1000)
        - 0.3 * freight_ratio
        + 0.05 * is_weekend
    )
    satisfaction_prob = np.clip(satisfaction_prob, 0.05, 0.95)
    satisfied = np.random.binomial(1, satisfaction_prob)
    
    df = pd.DataFrame({
        'delivery_days': delivery_days.round(1),
        'price': price.round(2),
        'freight_ratio': freight_ratio.round(3),
        'weight_kg': weight_kg.round(2),
        'is_weekend': is_weekend,
        'satisfied': satisfied
    })
    
    print(f"\n  Dataset: {len(df)} samples")
    print(f"  Satisfied: {satisfied.sum()} ({satisfied.mean()*100:.1f}%)")
    print(f"  Not satisfied: {n - satisfied.sum()} ({(1-satisfied.mean())*100:.1f}%)")
    
    # Train/test split
    X = df.drop('satisfied', axis=1)
    y = df['satisfied']
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
    
    # Train Decision Tree
    for criterion, name in [('entropy', 'ID3-style (Entropy)'), ('gini', 'CART (Gini)')]:
        print(f"\n  --- {name} ---")
        
        tree = DecisionTreeClassifier(
            criterion=criterion,
            max_depth=4,
            min_samples_leaf=20,
            random_state=42
        )
        tree.fit(X_train, y_train)
        
        y_pred = tree.predict(X_test)
        accuracy = accuracy_score(y_test, y_pred)
        
        print(f"  Accuracy: {accuracy:.4f}")
        print(f"\n  Classification Report:")
        print(classification_report(y_test, y_pred, target_names=['Not Satisfied', 'Satisfied']))
        
        # Feature importance
        importance = pd.Series(tree.feature_importances_, index=X.columns).sort_values(ascending=False)
        print(f"  Feature Importance:")
        for feat, imp in importance.items():
            bar = '█' * int(imp * 50)
            print(f"    {feat:<20} {imp:.4f} {bar}")
        
        # Print tree rules
        if criterion == 'entropy':
            print(f"\n  Decision Tree Rules (depth ≤ 3):")
            tree_rules = export_text(tree, feature_names=list(X.columns), max_depth=3)
            for line in tree_rules.split('\n')[:20]:
                print(f"    {line}")


# ==============================================================================
# PHẦN 3: K-MEANS CUSTOMER SEGMENTATION
# ==============================================================================

def demo_kmeans():
    """K-Means clustering cho RFM customer segmentation."""
    
    from sklearn.cluster import KMeans
    from sklearn.preprocessing import StandardScaler
    from sklearn.metrics import silhouette_score
    
    print("\n" + "=" * 70)
    print("  PART 3: K-MEANS - CUSTOMER SEGMENTATION (RFM)")
    print("=" * 70)
    
    np.random.seed(42)
    
    # Simulate RFM data
    n_customers = 500
    
    # Create clusters with different RFM profiles
    clusters_data = []
    
    # Champions: Recent, Frequent, High monetary
    for _ in range(100):
        clusters_data.append([
            np.random.uniform(1, 30),       # Recency (days)
            np.random.randint(5, 15),       # Frequency
            np.random.uniform(500, 2000),   # Monetary
        ])
    
    # Loyal: Moderate recency, frequent, moderate spend
    for _ in range(120):
        clusters_data.append([
            np.random.uniform(20, 90),
            np.random.randint(3, 10),
            np.random.uniform(200, 800),
        ])
    
    # New customers: Very recent, low frequency
    for _ in range(80):
        clusters_data.append([
            np.random.uniform(1, 15),
            np.random.randint(1, 3),
            np.random.uniform(50, 300),
        ])
    
    # At Risk: Old, used to be frequent
    for _ in range(100):
        clusters_data.append([
            np.random.uniform(90, 200),
            np.random.randint(3, 8),
            np.random.uniform(300, 1000),
        ])
    
    # Lost: Very old, infrequent, low spend
    for _ in range(100):
        clusters_data.append([
            np.random.uniform(150, 365),
            np.random.randint(1, 3),
            np.random.uniform(30, 200),
        ])
    
    rfm = pd.DataFrame(clusters_data, columns=['Recency', 'Frequency', 'Monetary'])
    
    print(f"\n  RFM Dataset: {len(rfm)} customers")
    print(f"\n  Statistics:")
    print(rfm.describe().round(2).to_string())
    
    # Standardize
    scaler = StandardScaler()
    rfm_scaled = scaler.fit_transform(rfm)
    
    # Elbow method
    print(f"\n  --- Elbow Method ---")
    inertias = []
    silhouettes = []
    K_range = range(2, 10)
    
    for k in K_range:
        km = KMeans(n_clusters=k, random_state=42, n_init=10)
        km.fit(rfm_scaled)
        inertias.append(km.inertia_)
        sil = silhouette_score(rfm_scaled, km.labels_)
        silhouettes.append(sil)
        print(f"    K={k}: Inertia={km.inertia_:.0f}, Silhouette={sil:.4f}")
    
    best_k = list(K_range)[np.argmax(silhouettes)]
    print(f"\n  Best K by Silhouette: {best_k}")
    
    # Final model with K=5
    K = 5
    km_final = KMeans(n_clusters=K, random_state=42, n_init=10)
    rfm['Cluster'] = km_final.fit_predict(rfm_scaled)
    
    # Cluster profiles
    print(f"\n  --- Cluster Profiles (K={K}) ---")
    profiles = rfm.groupby('Cluster').agg({
        'Recency': 'mean',
        'Frequency': 'mean',
        'Monetary': 'mean',
        'Cluster': 'count'
    }).rename(columns={'Cluster': 'Count'}).round(1)
    
    # Name clusters
    segment_names = {}
    for idx, row in profiles.iterrows():
        if row['Recency'] < 30 and row['Frequency'] > 5:
            segment_names[idx] = 'Champions'
        elif row['Recency'] < 60 and row['Frequency'] > 3:
            segment_names[idx] = 'Loyal'
        elif row['Recency'] < 30 and row['Frequency'] <= 2:
            segment_names[idx] = 'New Customers'
        elif row['Recency'] > 100 and row['Frequency'] > 3:
            segment_names[idx] = 'At Risk'
        else:
            segment_names[idx] = 'Lost/Hibernating'
    
    profiles['Segment'] = profiles.index.map(segment_names)
    print(profiles.to_string())
    
    # Visualization
    fig, axes = plt.subplots(1, 3, figsize=(16, 5))
    
    # Elbow
    axes[0].plot(list(K_range), inertias, 'bo-')
    axes[0].set_xlabel('K (Number of Clusters)')
    axes[0].set_ylabel('Inertia')
    axes[0].set_title('Elbow Method')
    
    # Silhouette
    axes[1].plot(list(K_range), silhouettes, 'ro-')
    axes[1].set_xlabel('K')
    axes[1].set_ylabel('Silhouette Score')
    axes[1].set_title('Silhouette Method')
    axes[1].axvline(x=best_k, color='green', linestyle='--', label=f'Best K={best_k}')
    axes[1].legend()
    
    # Scatter: Recency vs Monetary (colored by cluster)
    colors = ['#e74c3c', '#3498db', '#2ecc71', '#f39c12', '#9b59b6']
    for cluster in range(K):
        mask = rfm['Cluster'] == cluster
        name = segment_names.get(cluster, f'Cluster {cluster}')
        axes[2].scatter(rfm.loc[mask, 'Recency'], rfm.loc[mask, 'Monetary'],
                       c=colors[cluster % len(colors)], label=name, alpha=0.6, s=30)
    axes[2].set_xlabel('Recency (days)')
    axes[2].set_ylabel('Monetary ($)')
    axes[2].set_title(f'Customer Segments (K={K})')
    axes[2].legend(fontsize=8)
    
    plt.tight_layout()
    plt.savefig('kmeans_segmentation.png', dpi=150, bbox_inches='tight')
    print(f"\n  [OK] Saved: kmeans_segmentation.png")
    plt.close()


# ==============================================================================
# PHẦN 4: DBSCAN ANOMALY DETECTION
# ==============================================================================

def demo_dbscan():
    """DBSCAN cho phát hiện anomaly trong delivery data."""
    
    from sklearn.cluster import DBSCAN
    from sklearn.preprocessing import StandardScaler
    
    print("\n" + "=" * 70)
    print("  PART 4: DBSCAN - ANOMALY DETECTION")
    print("=" * 70)
    
    np.random.seed(42)
    
    # Simulate order data with anomalies
    n_normal = 400
    n_anomaly = 30
    
    # Normal orders
    normal = pd.DataFrame({
        'delivery_days': np.random.normal(12, 3, n_normal),
        'price': np.random.lognormal(4.5, 0.5, n_normal),
        'freight_ratio': np.random.normal(0.15, 0.05, n_normal),
    })
    
    # Anomalies
    anomalies = pd.DataFrame({
        'delivery_days': np.concatenate([
            np.random.uniform(40, 80, 15),   # Very late deliveries
            np.random.uniform(0, 1, 15),     # Suspiciously fast
        ]),
        'price': np.concatenate([
            np.random.uniform(2000, 10000, 15),  # Very expensive
            np.random.uniform(0.5, 5, 15),       # Suspiciously cheap
        ]),
        'freight_ratio': np.concatenate([
            np.random.uniform(0.5, 1.5, 15),  # High freight
            np.random.uniform(0, 0.01, 15),   # Almost no freight
        ]),
    })
    
    df = pd.concat([normal, anomalies], ignore_index=True)
    df['is_anomaly_truth'] = [0] * n_normal + [1] * n_anomaly
    
    # Standardize
    features = ['delivery_days', 'price', 'freight_ratio']
    scaler = StandardScaler()
    X_scaled = scaler.fit_transform(df[features])
    
    # DBSCAN
    # Try different eps values
    print(f"\n  Dataset: {len(df)} orders ({n_anomaly} known anomalies)")
    print(f"\n  --- DBSCAN Parameter Search ---")
    
    best_eps = 0.5
    best_score = 0
    
    for eps in [0.3, 0.5, 0.7, 1.0, 1.5]:
        for min_samples in [3, 5, 10]:
            db = DBSCAN(eps=eps, min_samples=min_samples)
            labels = db.fit_predict(X_scaled)
            
            n_clusters = len(set(labels)) - (1 if -1 in labels else 0)
            n_noise = (labels == -1).sum()
            
            # Check overlap with true anomalies
            detected_anomalies = set(np.where(labels == -1)[0])
            true_anomalies = set(np.where(df['is_anomaly_truth'] == 1)[0])
            
            true_positives = len(detected_anomalies & true_anomalies)
            precision = true_positives / len(detected_anomalies) if detected_anomalies else 0
            recall = true_positives / len(true_anomalies) if true_anomalies else 0
            f1 = 2 * precision * recall / (precision + recall) if (precision + recall) > 0 else 0
            
            print(f"    eps={eps}, min_samples={min_samples}: "
                  f"clusters={n_clusters}, noise={n_noise}, "
                  f"P={precision:.2f}, R={recall:.2f}, F1={f1:.2f}")
            
            if f1 > best_score:
                best_score = f1
                best_eps = eps
                best_min = min_samples
    
    # Best model
    print(f"\n  Best: eps={best_eps}, min_samples={best_min}, F1={best_score:.2f}")
    
    db_best = DBSCAN(eps=best_eps, min_samples=best_min)
    df['dbscan_label'] = db_best.fit_predict(X_scaled)
    df['detected_anomaly'] = (df['dbscan_label'] == -1).astype(int)
    
    # Print anomalies
    anomalies_detected = df[df['detected_anomaly'] == 1]
    print(f"\n  Detected {len(anomalies_detected)} anomalies:")
    print(anomalies_detected[features + ['is_anomaly_truth']].describe().round(2).to_string())
    
    # Visualization
    fig, axes = plt.subplots(1, 2, figsize=(14, 5))
    
    # Delivery days vs Price
    for label, name, color, marker in [
        (0, 'Normal', '#3498db', 'o'),
        (1, 'Anomaly (DBSCAN)', '#e74c3c', 'x')
    ]:
        mask = df['detected_anomaly'] == label
        axes[0].scatter(df.loc[mask, 'delivery_days'], df.loc[mask, 'price'],
                       c=color, label=name, alpha=0.6, s=30 if label == 0 else 80,
                       marker=marker)
    axes[0].set_xlabel('Delivery Days')
    axes[0].set_ylabel('Price ($)')
    axes[0].set_title('DBSCAN Anomaly Detection')
    axes[0].legend()
    
    # Confusion-like comparison
    tp = ((df['detected_anomaly'] == 1) & (df['is_anomaly_truth'] == 1)).sum()
    fp = ((df['detected_anomaly'] == 1) & (df['is_anomaly_truth'] == 0)).sum()
    fn = ((df['detected_anomaly'] == 0) & (df['is_anomaly_truth'] == 1)).sum()
    tn = ((df['detected_anomaly'] == 0) & (df['is_anomaly_truth'] == 0)).sum()
    
    confusion = np.array([[tn, fp], [fn, tp]])
    im = axes[1].imshow(confusion, cmap='Blues', interpolation='nearest')
    axes[1].set_xticks([0, 1])
    axes[1].set_yticks([0, 1])
    axes[1].set_xticklabels(['Predicted\nNormal', 'Predicted\nAnomaly'])
    axes[1].set_yticklabels(['Actual\nNormal', 'Actual\nAnomaly'])
    axes[1].set_title('Confusion Matrix')
    
    for i in range(2):
        for j in range(2):
            axes[1].text(j, i, str(confusion[i, j]),
                        ha='center', va='center', fontsize=16, fontweight='bold')
    
    plt.tight_layout()
    plt.savefig('dbscan_anomaly.png', dpi=150, bbox_inches='tight')
    print(f"\n  [OK] Saved: dbscan_anomaly.png")
    plt.close()


# ==============================================================================
# PHẦN 5: NAIVE BAYES
# ==============================================================================

def demo_naive_bayes():
    """Naive Bayes cho review score prediction."""
    
    from sklearn.naive_bayes import GaussianNB
    from sklearn.model_selection import train_test_split
    from sklearn.metrics import classification_report, accuracy_score
    
    print("\n" + "=" * 70)
    print("  PART 5: NAIVE BAYES - REVIEW PREDICTION")
    print("=" * 70)
    
    np.random.seed(42)
    n = 1000
    
    # Simulate features
    delivery_days = np.random.exponential(10, n)
    price = np.random.lognormal(4, 1, n)
    freight_ratio = np.random.uniform(0.05, 0.5, n)
    photos_qty = np.random.randint(1, 8, n)
    description_length = np.random.randint(50, 2000, n)
    
    # Target: Good review (>= 4)
    prob = (
        0.7
        - 0.015 * np.clip(delivery_days, 0, 30)
        + 0.01 * photos_qty
        + 0.0001 * description_length
        - 0.2 * freight_ratio
    )
    prob = np.clip(prob, 0.1, 0.9)
    good_review = np.random.binomial(1, prob)
    
    df = pd.DataFrame({
        'delivery_days': delivery_days,
        'price': price,
        'freight_ratio': freight_ratio,
        'photos_qty': photos_qty,
        'description_length': description_length,
        'good_review': good_review
    })
    
    X = df.drop('good_review', axis=1)
    y = df['good_review']
    
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
    
    # Gaussian Naive Bayes
    gnb = GaussianNB()
    gnb.fit(X_train, y_train)
    y_pred = gnb.predict(X_test)
    
    print(f"\n  Gaussian Naive Bayes:")
    print(f"  Accuracy: {accuracy_score(y_test, y_pred):.4f}")
    print(f"\n  Classification Report:")
    print(classification_report(y_test, y_pred, target_names=['Bad Review', 'Good Review']))
    
    # Class priors
    print(f"  Class Priors: {gnb.class_prior_}")
    
    # Feature means per class
    print(f"\n  Feature Means by Class:")
    means = pd.DataFrame(gnb.theta_, columns=X.columns, index=['Bad Review', 'Good Review'])
    print(means.round(2).to_string())
    
    # Predict example
    example = pd.DataFrame({
        'delivery_days': [5, 25],
        'price': [100, 500],
        'freight_ratio': [0.1, 0.4],
        'photos_qty': [5, 1],
        'description_length': [500, 100]
    })
    
    probs = gnb.predict_proba(example)
    print(f"\n  Prediction Examples:")
    for i, (_, row) in enumerate(example.iterrows()):
        print(f"    Order: delivery={row['delivery_days']}d, price=${row['price']}, "
              f"freight_ratio={row['freight_ratio']}")
        print(f"    → P(Good)={probs[i][1]:.3f}, P(Bad)={probs[i][0]:.3f}, "
              f"Predicted: {'Good' if probs[i][1] > 0.5 else 'Bad'}")


# ==============================================================================
# MAIN
# ==============================================================================

if __name__ == '__main__':
    print("=" * 70)
    print("  LAB 5: DATA MINING ALGORITHMS")
    print("  BIM5021 - Nha kho du lieu va Tich hop")
    print("=" * 70)
    
    demo_apriori()
    demo_decision_tree()
    demo_kmeans()
    demo_dbscan()
    demo_naive_bayes()
    
    print("\n" + "=" * 70)
    print("  HOAN THANH LAB 5!")
    print("  Files: kmeans_segmentation.png, dbscan_anomaly.png")
    print("=" * 70)