DLH — Machine Learning
A structured curriculum building the mathematical and data engineering foundations for machine learning.
Directory Structure
dlh-machine_learning/
├── math/ # Mathematical foundations
│ ├── linear_algebra/ # Matrix operations: Python lists → NumPy
│ │ ├── 0-slice_me_up.py through 14-saddle_up.py
│ │ ├── 100-slice_like_a_ninja.py through 102-squashed_like_sardines.py
│ │ └── README.md
│ ├── advanced_linear_algebra/ # Determinant, minor, cofactor, adjugate, inverse, definiteness
│ │ ├── 0-determinant.py through 5-definiteness.py, quiz.py
│ │ └── README.md
│ ├── calculus/ # Derivatives, integrals, partial derivatives, double integrals
│ │ ├── 0-sigma_is_for_sum through 17-integrate.py
│ │ └── README.md
│ ├── bayesian_prob/ # Likelihood, intersection, marginal, posterior probability
│ │ ├── 0-likelihood.py through 3-posterior.py
│ │ └── README.md
│ ├── plotting/ # Matplotlib: line, scatter, bar, frequency, PCA, gradient
│ │ ├── 0-line.py through 101-pca.py
│ │ └── README.md
│ ├── probability/ # Distributions: binomial, normal, poisson, exponential
│ │ ├── binomial.py, normal.py, poisson.py, exponential.py
│ │ └── README.md
│ ├── multivariate_prob/ # Mean vector, covariance, correlation, multivariate normal
│ │ ├── 0-mean_cov.py, 1-correlation.py, multinormal.py
│ │ └── README.md
│ └── README.md
├── pipeline/ # Data engineering
│ ├── databases/ # SQL: creation, CRUD, joins, aggregates, triggers
│ │ ├── 0-create_database_if_missing.sql through 18-valid_email.sql
│ │ ├── hbtn_0d_tvshows.sql, hbtn_0d_tvshows_rate.sql
│ │ ├── metal_bands.sql, temperatures.sql
│ │ └── README.md
│ ├── pandas/ # Pandas: DataFrame creation, cleaning, concat, resample, plot
│ │ ├── 0-from_numpy.py through 14-visualize.py
│ │ ├── *.csv, *.csv.zip
│ │ └── README.md
│ └── README.md
├── supervised_learning/ # Neural networks & deep learning
│ ├── classification/ # Neuron → neural network → deep NN (NumPy)
│ │ ├── 0-neuron.py through 15-neural_network.py
│ │ ├── README.md, RESOURCES.md
│ │ └── data/ # MNIST + binary datasets (gitignored)
│ └── keras/ # TensorFlow 2 & Keras
│ ├── 0-sequential.py through 10-weights.py
│ └── README.md, RESOURCES.md
├── unsupervised_learning/ # Clustering & dimensionality reduction
│ ├── clustering/ # k-means, GMM, agglomerative (NumPy)
│ │ ├── 0-initialize.py through 12-agglomerative.py
│ │ └── README.md, RESOURCES.md
│ └── dimensionality_reduction/ # PCA, t-SNE (NumPy)
│ ├── 0-pca.py through 8-tsne.py
│ └── README.md, RESOURCES.md
├── my-venv/ # Python virtual environment
└── README.md
Quick Reference
| Track | Module | Topics | Tasks |
|---|---|---|---|
| Math | Linear Algebra | Slicing, shape, transpose, element-wise ops, concat, matrix multiply, NumPy, n-D generalization | 19 |
| Math | Advanced Linear Algebra | Determinant, minor, cofactor, adjugate, inverse, definiteness (manual + NumPy) | 7 |
| Math | Calculus | Derivatives, partial derivatives, integrals, definite/indefinite, double integrals | 17 |
| Math | Bayesian Probability | Likelihood, intersection, marginal, posterior probability | 4 |
| Math | Plotting | Line, scatter, bar, frequency, all-in-one, gradient descent, PCA | 9 |
| Math | Probability | Binomial, normal, poisson, exponential distributions | 4 |
| Math | Multivariate Probability | Mean vector, covariance, correlation, multivariate normal distribution | 3 |
| Pipeline | Databases | DDL, CRUD, WHERE, ORDER BY, GROUP BY, JOINS, aggregates, constraints, triggers | 18 (+4 schemas) |
| Pipeline | Pandas | DataFrame creation, rename, slice, fill, concat, hierarchy, describe, resample, visualize | 15 |
| Supervised | Classification | Binary classification: neuron (logistic), NN with one hidden layer, deep NN; NumPy from scratch | 16 |
| Supervised | Keras | TensorFlow 2 & Keras: Sequential, Input, optimization, training, model persistence | 11 |
| Unsupervised | Clustering | k-means, variance, optimum k, Gaussian Mixture Models, EM, BIC, agglomerative | 13 |
| Unsupervised | Dimensionality Reduction | PCA, t-SNE (P/affinities, gradient descent, cost) | 9 |
Learning Progression
Math Track
- Python Slicing → 2. Manual Matrix Ops (nested loops) → 3. NumPy Vectorization → 4. N-Dimensional Generalization → 5. Advanced Linear Algebra (determinant → inverse → definiteness) → 6. Calculus (derivatives → integrals) → 7. Probability & Statistics (distributions → Bayesian) → 8. Multivariate Probability (mean/cov → correlation → multivariate normal) → 9. Visualization (plotting → PCA)
Pipeline Track
- Foundation (CREATE) → 2. CRUD → 3. Filtering/Sorting → 4. Joins → 5. Constraints → 6. Real-World Data → 7. Triggers → 8. Pandas (DataFrame ops → cleaning → concat → resample → visualize)
Supervised Learning Track
- Classification Fundamentals (neuron → single hidden layer → deep NN, NumPy from scratch) → 2. Keras (high-level API)
Unsupervised Learning Track
- Clustering (k-means → variance → GMM/EM → agglomerative) → 2. Dimensionality Reduction (PCA → t-SNE)
Setup
cd dlh-machine_learning
source my-venv/bin/activate
pip install numpy
Keras code (supervised_learning/keras/) runs in tf-venv (Python 3.11, TensorFlow 2.15).