import os import random import zipfile from datetime import datetime, timedelta import numpy as np import pandas as pd BASE_DIR = os.path.dirname(os.path.abspath(__file__)) INDUSTRIES = [ 'Retail', 'Banking', 'Telecom', 'Healthcare', 'E-commerce', ] REGIONS = [ {'region': 'North', 'cities': ['Hà Nội', 'Hải Phòng', 'Bắc Ninh', 'Hải Dương']}, {'region': 'Central', 'cities': ['Đà Nẵng', 'Huế', 'Nha Trang', 'Quy Nhơn']}, {'region': 'South', 'cities': ['TP.HCM', 'Cần Thơ', 'Bình Dương', 'Đồng Nai']}, ] PRODUCT_CATEGORIES = [ ('Thực phẩm', 'Sữa & Đồ uống'), ('Thực phẩm', 'Snack & Đồ ăn nhanh'), ('Chăm sóc cá nhân', 'Dầu gội & Sữa tắm'), ('Mỹ phẩm', 'Trang điểm'), ('Điện tử', 'Tai nghe & Loa'), ('Thời trang', 'Giày dép'), ('Đồ gia dụng', 'Đồ bếp'), ] BRANDS = [ 'Vinamilk', 'TH True Milk', 'Unilever', 'P&G', 'Samsung', 'Apple', 'Xiaomi', 'OPPO', 'L’Oreal', 'Maybelline', 'Nestle', 'Coca-Cola', 'Kinh Đô', 'Lock&Lock', 'Panasonic', 'Sony', 'Adidas', 'Nike', 'JBL', 'Lifebuoy' ] CUSTOMER_SEGMENTS = ['Standard', 'Bronze', 'Silver', 'Gold', 'Platinum'] GENDERS = ['Nam', 'Nữ'] PAYMENT_METHODS = ['Cash', 'Credit Card', 'E-wallet', 'Bank Transfer'] WORKFLOW_STATUSES = ['success', 'failed', 'running'] SOCIAL_PLATFORMS = ['Facebook', 'Instagram', 'LinkedIn', 'TikTok', 'Email'] CAMPAIGN_OBJECTIVES = ['Brand Awareness', 'Lead Generation', 'Sales', 'Retention', 'Launch'] CONTENT_TONES = ['Chuyên nghiệp', 'Thân thiện', 'Hấp dẫn', 'Thuyết phục', 'Giải trí'] def make_customers(n_customers=5000): customers = [] for i in range(1, n_customers + 1): region_info = random.choice(REGIONS) city = random.choice(region_info['cities']) age = random.randint(18, 65) industry = random.choice(INDUSTRIES) segment = random.choices(CUSTOMER_SEGMENTS, weights=[45, 25, 15, 10, 5], k=1)[0] registration_date = datetime(2021, 1, 1) + timedelta(days=random.randint(0, 730)) first_purchase_date = registration_date + timedelta(days=random.randint(0, 90)) customers.append({ 'customer_id': f'CUST{str(i).zfill(5)}', 'full_name': f'Khách hàng {i}', 'gender': random.choice(GENDERS), 'age': age, 'age_group': f'{(age // 10) * 10}-{(age // 10) * 10 + 9}', 'phone_prefix': random.choice(['090', '091', '092', '093', '094', '095', '096', '097', '098', '099']), 'email_domain': random.choice(['gmail.com', 'yahoo.com', 'vng.com', 'fpt.com', 'vnpt.vn']), 'city': city, 'region': region_info['region'], 'industry': industry, 'customer_segment': segment, 'loyalty_points': random.randint(0, 20000), 'registration_date': registration_date.date(), 'first_purchase_date': first_purchase_date.date(), 'is_active': random.choices([True, False], weights=[0.85, 0.15], k=1)[0], }) return pd.DataFrame(customers) def make_products(n_products=500): products = [] for i in range(1, n_products + 1): category, subcategory = random.choice(PRODUCT_CATEGORIES) brand = random.choice(BRANDS) industry = random.choice(INDUSTRIES) unit_cost = random.randint(10000, 200000) unit_price = int(unit_cost * random.uniform(1.1, 2.5)) products.append({ 'product_id': f'PROD{str(i).zfill(5)}', 'product_name': f'{brand} {subcategory} {i}', 'category': category, 'subcategory': subcategory, 'brand': brand, 'industry': industry, 'unit_price': unit_price, 'unit_cost': unit_cost, 'profit_margin': round((unit_price - unit_cost) / unit_price * 100, 2), 'unit_of_measure': random.choice(['Cái', 'Hộp', 'Chai', 'Bộ', 'Chiếc']), 'country_origin': random.choice(['Việt Nam', 'Nhật Bản', 'Hàn Quốc', 'Mỹ', 'Trung Quốc']), 'launch_date': (datetime(2020, 1, 1) + timedelta(days=random.randint(0, 1500))).date(), 'is_active': True, }) return pd.DataFrame(products) def make_stores(n_stores=50): stores = [] for i in range(1, n_stores + 1): region_info = random.choice(REGIONS) city = random.choice(region_info['cities']) stores.append({ 'store_id': f'ST{str(i).zfill(3)}', 'store_name': f'Chi nhánh {city} {i}', 'region': region_info['region'], 'city': city, 'store_type': random.choice(['Store', 'Branch', 'Outlet', 'Mini-store']), 'opened_date': (datetime(2018, 1, 1) + timedelta(days=random.randint(0, 2000))).date(), 'is_flagship': random.choice([True, False]), }) return pd.DataFrame(stores) def make_transactions(customers, products, stores, n_transactions=50000): transactions = [] start_date = datetime(2023, 1, 1) for i in range(1, n_transactions + 1): customer = customers.sample(1).iloc[0] product = products.sample(1).iloc[0] store = stores.sample(1).iloc[0] transaction_date = start_date + timedelta(days=random.randint(0, 729)) quantity = random.randint(1, 10) discount_percent = random.choice([0, 0, 5, 10, 15]) total_amount = int(product['unit_price'] * quantity * (1 - discount_percent / 100)) transactions.append({ 'transaction_id': f'TXN{str(i).zfill(6)}', 'customer_id': customer['customer_id'], 'product_id': product['product_id'], 'store_id': store['store_id'], 'transaction_date': transaction_date.date(), 'industry': random.choice([customer['industry'], product['industry'], store['region']]), 'region': store['region'], 'city': store['city'], 'quantity': quantity, 'unit_price': product['unit_price'], 'discount_percent': discount_percent, 'total_amount': total_amount, 'payment_method': random.choice(PAYMENT_METHODS), 'is_returned': random.choices([False, True], weights=[0.95, 0.05], k=1)[0], }) return pd.DataFrame(transactions) def save_and_zip(dataframes, folder_name, zip_name): output_dir = os.path.join(BASE_DIR, folder_name) os.makedirs(output_dir, exist_ok=True) for filename, df in dataframes.items(): df.to_csv(os.path.join(output_dir, filename), index=False) zip_path = os.path.join(BASE_DIR, zip_name) with zipfile.ZipFile(zip_path, 'w', compression=zipfile.ZIP_DEFLATED) as archive: for filename in dataframes.keys(): archive.write(os.path.join(output_dir, filename), arcname=os.path.join(folder_name, filename)) print(f'Generated {zip_path}') return output_dir, zip_path def make_business_analytics_data(): customers = make_customers(n_customers=2000) products = make_products(n_products=300) sales = [] start_date = datetime(2023, 1, 1) for i in range(1, 15001): customer = customers.sample(1).iloc[0] product = products.sample(1).iloc[0] transaction_date = start_date + timedelta(days=random.randint(0, 729)) quantity = random.randint(1, 10) discount_percent = random.choice([0, 0, 5, 10]) total_amount = int(product['unit_price'] * quantity * (1 - discount_percent / 100)) sales.append({ 'sale_id': f'SALE{str(i).zfill(6)}', 'customer_id': customer['customer_id'], 'product_id': product['product_id'], 'transaction_date': transaction_date.date(), 'region': customer['region'], 'city': customer['city'], 'industry': customer['industry'], 'quantity': quantity, 'unit_price': product['unit_price'], 'discount_percent': discount_percent, 'total_amount': total_amount, 'payment_method': random.choice(PAYMENT_METHODS), }) sales_df = pd.DataFrame(sales) return { 'customers.csv': customers, 'products.csv': products, 'sales.csv': sales_df, } def make_ml_playground_data(): churn = [] for i in range(1, 8001): age = random.randint(18, 80) tenure = random.randint(1, 72) monthly = random.randint(200, 3000) churn.append({ 'customer_id': f'CHURN{str(i).zfill(5)}', 'age': age, 'gender': random.choice(GENDERS), 'tenure_months': tenure, 'monthly_charges': monthly, 'total_charges': round(monthly * tenure * random.uniform(0.9, 1.1), 2), 'churn': random.choices([0, 1], weights=[0.75, 0.25], k=1)[0], }) houses = [] locations = ['Hà Nội', 'TP.HCM', 'Đà Nẵng', 'Nha Trang', 'Huế', 'Cần Thơ'] for i in range(1, 5001): bedrooms = random.randint(1, 5) bathrooms = random.randint(1, 3) area = random.randint(30, 250) year = random.randint(1990, 2024) price = int(area * random.randint(2000, 6000) * (1 + (bedrooms - 1) * 0.1)) houses.append({ 'house_id': f'HOUSE{str(i).zfill(5)}', 'location': random.choice(locations), 'bedrooms': bedrooms, 'bathrooms': bathrooms, 'area_sqft': area, 'year_built': year, 'price': price, }) reviews = [] sentiments = ['positive', 'neutral', 'negative'] topics = ['electronics', 'food', 'fashion', 'travel', 'health'] for i in range(1, 12001): sentiment = random.choices(sentiments, weights=[0.55, 0.25, 0.2], k=1)[0] reviews.append({ 'review_id': f'REV{str(i).zfill(5)}', 'product_id': f'PROD{str(random.randint(1, 300)).zfill(5)}', 'rating': random.randint(1, 5), 'review_text': f'This {random.choice(topics)} product is {"great" if sentiment == "positive" else "okay" if sentiment == "neutral" else "disappointing"}.', 'review_date': (datetime(2023, 1, 1) + timedelta(days=random.randint(0, 729))).date(), 'sentiment': sentiment, }) return { 'customer_churn.csv': pd.DataFrame(churn), 'house_prices.csv': pd.DataFrame(houses), 'product_reviews.csv': pd.DataFrame(reviews), } def make_genai_document_corpus_data(): kb = [] domains = ['Sales', 'Support', 'HR', 'Finance', 'Product'] for i in range(1, 7501): domain = random.choice(domains) kb.append({ 'doc_id': f'KB{str(i).zfill(5)}', 'title': f'{domain} best practice {i}', 'content': f'Nội dung tài liệu {i} về {domain} với các quy trình và hướng dẫn.', 'domain': domain, 'source': random.choice(['Internal', 'Public Web', 'Partner', 'Research']), }) chat = [] roles = ['user', 'assistant'] for i in range(1, 5001): chat.append({ 'chat_id': f'CHAT{str(i).zfill(5)}', 'user_id': f'USER{str(random.randint(1, 2000)).zfill(4)}', 'timestamp': (datetime(2023, 1, 1) + timedelta(days=random.randint(0, 729), seconds=random.randint(0, 86399))).isoformat(), 'role': random.choice(roles), 'message': f'Message {i} về {random.choice(domains)}.', }) faq = [] for i in range(1, 1201): topic = random.choice(domains) faq.append({ 'faq_id': f'FAQ{str(i).zfill(4)}', 'question': f'Làm sao để giải quyết vấn đề {topic.lower()} {i}?', 'answer': f'Trả lời cho câu hỏi về {topic.lower()} trong ngữ cảnh doanh nghiệp.', 'topic': topic, }) return { 'knowledge_base.csv': pd.DataFrame(kb), 'chat_logs.csv': pd.DataFrame(chat), 'faq_pairs.csv': pd.DataFrame(faq), } def make_content_creator_library_data(): templates = [] industries = ['Marketing', 'Retail', 'Education', 'Finance', 'Travel'] for i in range(1, 1201): templates.append({ 'template_id': f'TEMP{str(i).zfill(5)}', 'content_type': random.choice(['Email', 'Post', 'Ad', 'Landing Page']), 'industry': random.choice(industries), 'tone': random.choice(CONTENT_TONES), 'template_text': f'Ví dụ template nội dung {i} cho {random.choice(industries)}.', }) briefs = [] for i in range(1, 801): briefs.append({ 'brief_id': f'BRIEF{str(i).zfill(4)}', 'campaign_name': f'Chiến dịch {i}', 'objective': random.choice(CAMPAIGN_OBJECTIVES), 'target_audience': random.choice(['Gen Z', 'Millennials', 'Doanh nghiệp vừa và nhỏ', 'Khách hàng cao cấp']), 'channel': random.choice(['Facebook', 'Instagram', 'Email', 'TikTok', 'LinkedIn']), 'budget': random.randint(5000000, 50000000), }) posts = [] for i in range(1, 1501): posts.append({ 'post_id': f'POST{str(i).zfill(4)}', 'platform': random.choice(SOCIAL_PLATFORMS), 'topic': random.choice(['Launch', 'Promotion', 'Storytelling', 'Tips', 'Event']), 'post_text': f'Nội dung bài đăng {i} cho {random.choice(SOCIAL_PLATFORMS)} về {random.choice(['xu hướng', 'khuyến mãi', 'sản phẩm'])}.', 'scheduled_date': (datetime(2024, 1, 1) + timedelta(days=random.randint(0, 365))).date(), }) return { 'content_templates.csv': pd.DataFrame(templates), 'campaign_briefs.csv': pd.DataFrame(briefs), 'social_posts.csv': pd.DataFrame(posts), } def make_automation_event_logs_data(): events = [] for i in range(1, 18001): workflow_id = f'WF{str(random.randint(1, 120)).zfill(3)}' start_time = datetime(2024, 1, 1) + timedelta(days=random.randint(0, 180), seconds=random.randint(0, 86399)) duration = random.randint(100, 20000) events.append({ 'event_id': f'EVENT{str(i).zfill(6)}', 'timestamp': start_time.isoformat(), 'workflow_id': workflow_id, 'step_name': random.choice(['start', 'validate', 'transform', 'send', 'complete']), 'status': random.choice(WORKFLOW_STATUSES), 'duration_ms': duration, 'payload_type': random.choice(['JSON', 'XML', 'CSV', 'Webhook']), 'error_code': random.choice(['', 'ERR001', 'ERR002', 'ERR_TIMEOUT']), }) connections = [] services = ['Salesforce', 'Google Sheets', 'Slack', 'Email', 'Stripe'] for i in range(1, 121): connections.append({ 'connection_id': f'CONN{str(i).zfill(4)}', 'service_name': random.choice(services), 'status': random.choice(['active', 'inactive']), 'last_tested': (datetime(2024, 1, 1) + timedelta(days=random.randint(0, 180))).date(), 'auth_type': random.choice(['API Key', 'OAuth2', 'Basic Auth']), 'region': random.choice(['Vietnam', 'APAC', 'Global']), }) task_runs = [] for i in range(1, 8501): start_at = datetime(2024, 1, 1) + timedelta(days=random.randint(0, 180), seconds=random.randint(0, 86399)) duration = random.randint(100, 15000) task_runs.append({ 'task_run_id': f'TASK{str(i).zfill(5)}', 'workflow_id': f'WF{str(random.randint(1, 120)).zfill(3)}', 'task_name': random.choice(['FetchData', 'Transform', 'SendNotification', 'Archive', 'Retry']), 'started_at': start_at.isoformat(), 'ended_at': (start_at + timedelta(milliseconds=duration)).isoformat(), 'status': random.choice(['success', 'failed', 'running']), 'attempt': random.randint(1, 3), 'duration_ms': duration, }) return { 'event_logs.csv': pd.DataFrame(events), 'integration_connections.csv': pd.DataFrame(connections), 'task_runs.csv': pd.DataFrame(task_runs), } def main(): dataset_generators = [ ('minai_business_analytics_starter_dataset', 'minai_business_analytics_starter_dataset.zip', make_business_analytics_data), ('minai_ml_playground_dataset', 'minai_ml_playground_dataset.zip', make_ml_playground_data), ('minai_genai_document_corpus_dataset', 'minai_genai_document_corpus_dataset.zip', make_genai_document_corpus_data), ('minai_content_creator_library_dataset', 'minai_content_creator_library_dataset.zip', make_content_creator_library_data), ('minai_automation_event_logs_dataset', 'minai_automation_event_logs_dataset.zip', make_automation_event_logs_data), ('minai_da_industry_dataset', 'minai_da_industry_dataset.zip', lambda: { 'customers.csv': make_customers(n_customers=5000), 'products.csv': make_products(n_products=500), 'stores.csv': make_stores(n_stores=50), 'transactions.csv': make_transactions(make_customers(n_customers=5000), make_products(n_products=500), make_stores(n_stores=50), n_transactions=50000), }), ] for folder_name, zip_name, generator in dataset_generators: print(f'Generating {folder_name}...') dataframes = generator() save_and_zip(dataframes, folder_name, zip_name) print('All datasets generated successfully.') if __name__ == '__main__': main()