Vincent777756/Analystlab-ml-internship

Week 1-2 | AnalystLab Africa ML Internship — EDA & preprocessing on Titanic (tabular) and IMDB Reviews (NLP) datasets. # AnalystLab Africa — Machine Learning Internship Documenting my weekly progress through the AnalystLab Africa Machine Learning Internship Program (Batch A: May 1 — July 1, 2026). --- ## 📁 Weekly Progress | Week | Topic | Folder | Status | |---|---|---|---| | Week 1-2 | Data Preprocessing & EDA | `week1-2-eda/` | Complete | | Week 3 | Machine Learning Fundamentals | `week3-ml-fundamentals/` | | Week 4 | Supervised Learning | `week4-supervised-learning/` | | Week 5 | Advanced Machine Learning | `week5-advanced-ml/` | | Week 6 | Model Tuning & Validation | `week6-model-tuning/` | | Week 7 | Model Deployment | `Week-7-deployment/` | | Week 8 | Capstone Project | `week8-capstone/` | --- ## Week 1-2: Data Preprocessing & EDA **Notebook:** EDA_Notebook.ipynb **Datasets:** Titanic | IMDB 50K Reviews Women on the Titanic survived at 74% vs 19% for men. IMDB dataset is perfectly balanced — 25,000 positive, 25,000 negative. --- ## Week 3: Machine Learning Fundamentals **Notebook:** Week3_ML_Fundamentals.ipynb | Model | Accuracy | |---|---| | Logistic Regression | 80.45% | | Random Forest | 81.56% | | IMDB Sentiment (TF-IDF + LR) | 86.40% | --- ## Week 4: Supervised Learning **Notebook:** Week4_Supervised_Learning.ipynb **Linear Regression (Boston Housing):** RMSE $5.14k, R² 0.64 **Logistic Regression (Titanic):** Accuracy 80.45% --- ## Week 5: Advanced Machine Learning **Notebook:** Week5_Advanced_ML.ipynb | Model | CV Accuracy | |---|---| | Decision Tree | 77.26% | | Random Forest | 78.39% | | Gradient Boosting | 81.19% | | RF Tuned (GridSearch) | 82.88% | --- ## Week 6: Model Tuning & Validation **Notebook:** Week6_Model_Tuning_Validation.ipynb **Dataset:** Pima Indians Diabetes Database | Model | CV Accuracy | |---|---| | Baseline (Random Forest) | 76.38% ± 2.14% | | Grid Search Tuned | 78.01% ± 2.58% | | Random Search Tuned | 78.18% ± 2.38% | Cross-validation accuracy improved after tuning even when single test-set accuracy dropped — CV is t …