golfbag Link to heading

status – finished linear regression (2025-09-17)

🏌️ 12-Week ML + ETL Curriculum (Golf Bag Training) Link to heading

This curriculum covers 12 canonical ML categories (the “golf bag”) with ETL practice.
Each week introduces a dataset, ETL tasks, and ML goals. Use this as a checklist in Obsidian.

📅 Week 1: Linear Regression Link to heading

• Dataset: California Housing (scikit-learn)
• ETL Tasks:
  • Load CSV, validate schema
  • Handle missing values
  • Normalize numeric fields
• ML Tasks:
  • Train/test split
  • Fit linear regression
  • Interpret coefficients

📅 Week 2: Logistic Regression Link to heading

• Dataset: Breast Cancer Wisconsin
• ETL Tasks:
  • Encode categorical features
  • Scale inputs
• ML Tasks:
  • Train binary classifier
  • Evaluate with ROC, precision, recall

📅 Week 3: Decision Trees Link to heading

• Dataset: Iris Dataset
• ETL Tasks:
  • Basic EDA (histograms, scatter plots)
  • Verify balanced classes
• ML Tasks:
  • Build decision tree
  • Visualize splits
  • Observe overfitting

📅 Week 4: Random Forests Link to heading

• Dataset: UCI Adult / Census Income
• ETL Tasks:
  • Encode categorical features
  • Impute missing values
  • Balance classes
• ML Tasks:
  • Train random forest
  • Measure feature importance
  • Compare to logistic regression

📅 Week 5: K-Means Clustering Link to heading

• Dataset: MNIST Digits (scikit-learn or full MNIST)
• ETL Tasks:
  • Flatten images → tabular form
  • Apply PCA for preprocessing
• ML Tasks:
  • Run K-Means (k=10)
  • Visualize clusters
  • Evaluate with silhouette score

📅 Week 6: PCA (Dimensionality Reduction) Link to heading

• Dataset: UCI Wine Dataset
• ETL Tasks:
  • Normalize features
  • Check correlations
• ML Tasks:
  • Run PCA
  • Plot explained variance
  • Create 2D scatter plot

📅 Week 7: Neural Networks (MLP) Link to heading

• Dataset: MNIST Digits
• ETL Tasks:
  • Normalize pixel intensities
  • Train/val/test split
  • Manage batch sizes
• ML Tasks:
  • Build 2-layer MLP
  • Compare accuracy to logistic regression

📅 Week 8: CNN (Convolutions) Link to heading

• Dataset: CIFAR-10
• ETL Tasks:
  • Resize and augment images
  • Split train/val/test
• ML Tasks:
  • Build CNN with conv + pooling
  • Apply dropout
  • Compare to transfer learning

📅 Week 9: RNN / LSTM (Sequences) Link to heading

• Dataset: Shakespeare Corpus (tiny) or IMDB Reviews
• ETL Tasks:
  • Tokenize sequences
  • Pad to fixed length
  • Manage vocabulary size
• ML Tasks:
  • Train LSTM
  • Predict next char / classify sentiment

📅 Week 10: Transformers (BERT) Link to heading

• Dataset: SST-2 Sentiment Treebank
• ETL Tasks:
  • Tokenize with HuggingFace
  • Create train/dev/test splits
• ML Tasks:
  • Fine-tune BERT
  • Evaluate F1 score
  • Compare to LSTM baseline

📅 Week 11: Reinforcement Learning (Q-Learning) Link to heading

• Dataset/Env: OpenAI Gym CartPole
• ETL Tasks:
  • Log state/action/reward data
  • Define schema for episodes
• ML Tasks:
  • Train Q-learning agent
  • Plot reward curve
  • Test convergence

📅 Week 12: Anomaly Detection Link to heading

• Dataset: Kaggle Credit Card Fraud Detection
• ETL Tasks:
  • Handle extreme class imbalance
  • Scale features
  • Stratified train/test split
• ML Tasks:
  • Train Isolation Forest
  • Compare to Autoencoder
  • Evaluate precision@k

📝 Notes Link to heading

• Each week: spend equal time on ETL and ML — good pipelines make models shine.
• Keep notebooks + scripts versioned (Git).
• Optional stretch goal: orchestrate pipelines (Airflow, Dagster, Prefect).
• Track experiments: MLflow, Weights & Biases, or simple CSV logs.