Kapsamlı Data Science ve MLOps eğitimi - 10 haftalık program
Konular: DVC, MLflow, FastAPI, Docker, GitHub Actions
- 📁 01-dvc-versioning - Veri versiyonlama
- 📁 02-mlflow-tracking - Deney takibi
- 📁 03-fastapi-ml-api - ML API geliştirme
- 📁 04-docker-deployment - Containerization
- 📁 05-github-actions-ci - CI/CD pipeline
- 🚢 titanic-mlops - Kapsamlı proje örneği
Konular: SMOTE, Undersampling, Class Weights, Logistic Regression vs XGBoost, Streamlit Dashboard
- 💰 loan-risk-analysis - Kredi risk analizi projesi
- Kaggle veri seti entegrasyonu
- SMOTE ve undersampling teknikleri
- Model karşılaştırması (LR vs XGBoost)
- Streamlit dashboard geliştirme
- 🚀 Render deployment ready
Konular: Outlier Detection, Feature Engineering, ROC-AUC/PR-AUC, SHAP/LIME, CI/CD Pipeline
- 📁 modul - Eğitim modülleri (5 script)
- Isolation Forest ve LOF outlier detection
- Feature scaling ve encoding yöntemleri
- ROC-AUC ve PR-AUC metrikleri analizi
- SHAP/LIME model açıklanabilirlik
- CI/CD pipeline ve deployment
- 🏦 fraud-detection - Kapsamlı fraud detection projesi
- Credit Card Fraud Detection dataset
- Gerçek dünya imbalanced learning
- Production-ready MLOps pipeline
- Interactive demo ve comprehensive evaluation
Konular: ARIMA, Prophet, LightGBM, Time Series CV, Prefect Automation, Docker Deployment
- 🏪 hafta5 - M5 Forecasting Modular Pipeline
- 📖 M5 Dataset Story: Walmart'ın 5 yıllık satış verisi (2011-2016)
- 🏗️ Modular Architecture: 7 specialized modules (P1-P7)
- 📊 Model Comparison: ARIMA (~46% sMAPE) vs Prophet (~28% sMAPE) vs LightGBM (~33% sMAPE)
- ⚙️ Feature Engineering: Lag, rolling, seasonal features
- ✅ Time Series CV: Rolling-origin cross-validation (proper temporal splitting)
- 🔄 Prefect Automation: Daily forecasting pipeline (09:00 Europe/Istanbul)
- 🐳 Docker Ready: Production containerization with volume mounting
- 📚 Comprehensive Docs: 1,500+ line documentation with dataset story
Konular: Association Rules, Support-Confidence-Lift, Cross-selling, Interactive Analysis
- 🛒 hafta6 - Market Sepeti Analizi Projesi
- Market Basket Analysis temelleri
- Support, Confidence, Lift kavramları
- Association Rules ve ürün birliktelikleri
- Cross-selling ve mağaza optimizasyonu
- Interactive Streamlit dashboard
- 📊 999 sepet x 16 ürün veri seti analizi
Konular: Text Cleaning, Tokenization, TF-IDF, BERT, PII Masking, Model Optimization, FastAPI
- 📝 hafta7 - Türkçe NLP ve Model Optimizasyonu Projesi
- Sentetik sağlık verisi (500 kayıt) ile çalışma
- Türkçe metin temizleme ve tokenization (NLTK + SpaCy)
- TF-IDF analizi ve doküman benzerliği
- BERT tabanlı semantik analiz (multilingual)
- PII maskeleme (Regex + Presidio) ile güvenlik
- Model optimizasyonu (DistilBERT, quantization)
- Production-ready FastAPI servisi
- 🏥 Sağlık verilerinde gizlilik koruması
- İçerik belirlenecek
- İçerik belirlenecek
- Proje sunumları ve değerlendirme
# Proje klasörüne git
cd hafta7/
# Virtual environment aktifleştir
source venv/bin/activate
# SpaCy modelini yükle
python -m spacy download en_core_web_sm
# Tüm örnekleri çalıştır (demo)
python run_examples.py
# API servisi başlat
python api/main.py # Terminal 1
python api/test_api.py # Terminal 2 - API testleri
# PII maskeleme örneği
python src/pii_masking.py# Proje klasörüne git
cd hafta6/
# Virtual environment aktifleştir
source venv/bin/activate
# Konsol uygulaması çalıştır
python basit_market_analizi.py
# Web dashboard başlat
streamlit run basit_streamlit_app.py# Proje klasörüne git
cd hafta4/fraud-detection/
# Virtual environment oluştur
python -m venv venv && source venv/bin/activate
# Dependencies kur
pip install -r requirements.txt
# Interactive demo başlat
python run_demo.py
# Eğitim modülleri çalıştır
cd ../modul/
python 1_outlier_detection_with_save.py# Proje klasörüne git
cd hafta3/loan-risk-analysis/
# Kurulumu başlat
./start.sh
# EDA analizi yap
cd src && python eda.py
# Model eğitimi
python models.py
# Streamlit dashboard
streamlit run streamlit_app/app.py# Sanal ortamı aktif et
source ds360/bin/activate
# İkinci hafta projelerine git
cd hafta2/
# Titanic MLOps projesini incele
cd titanic-mlops/
uvicorn src.api:app --reload- DVC → Veri versiyonlama temelleri
- MLflow → Model eğitimi ve takip
- FastAPI → API geliştirme
- Docker → Containerization
- GitHub Actions → CI/CD automation
- Titanic MLOps → Tüm teknolojilerin entegrasyonu
- EDA → Veri keşfi ve analizi
- SMOTE → Sentetik örnekleme
- Undersampling → Çoğunluk sınıf azaltma
- Class Weights → Sınıf ağırlıklandırma
- Model Comparison → LR vs XGBoost
- Streamlit → Dashboard ve deployment
- Outlier Detection → Isolation Forest & LOF
- Feature Engineering → Scaling, encoding, imbalance handling
- Advanced Metrics → ROC-AUC vs PR-AUC analysis
- Explainability → SHAP & LIME model interpretation
- CI/CD Pipeline → Production deployment strategies
- Business Impact → Cost-benefit analysis & threshold optimization
- Data Understanding → Transaction data and product relationships
- Support Calculation → Item frequency analysis
- Association Rules → Confidence and lift metrics
- Interactive Analysis → Streamlit dashboard development
- Business Applications → Cross-selling and store optimization
- Advanced Patterns → Multi-item combinations and recommendations
- Text Preprocessing → Turkish text cleaning and tokenization
- TF-IDF Analysis → Term frequency and document similarity
- BERT Integration → Semantic analysis with multilingual models
- PII Protection → Privacy preservation with regex and AI
- Model Optimization → DistilBERT and quantization techniques
- API Development → Production FastAPI services
- KAPSAMLI_REHBER.md - Detaylı teknik rehber
- 🚀 RENDER_DEPLOYMENT_GUIDE.md - Multi-project Render deployment
- Her proje klasöründeki README.md dosyalarını okuyun
- Practical örnekler ve hands-on projeler
Bu bootcamp MLOps dünyasında profesyonel olmak için gerekli tüm becerileri kapsar