irachrist1/Mathematics-for-Machine-Learning-PCA_Formative

PCA from scratch on African health data — ALU formative # Mathematics-for-Machine-Learning-PCA_Formative PCA from scratch on African malaria and health indicators — ALU Mathematics for Machine Learning formative assignment. ## The problem You have 27 columns of health data across 594 African countries. Plotting everything at once is noise. You need dimensionality reduction that preserves variance — implemented yourself in NumPy, not `sklearn.decomposition` black box. ## What it does Implements PCA manually: center data, compute covariance, eigendecomposition, project to principal components. Analyzes `DatasetAfricaMalaria.csv` from Kaggle. ``` 594 rows × 27 columns → cleaned 15 features → 2–3 principal components explaining most variance ``` ## Install ```bash git clone github.com && cd Mathematics-for-Machine-Learning-PCA_Formative pip install numpy pandas matplotlib jupyter jupyter notebook PCA_Formative_Gentil_Iradukunda.ipynb ``` ## How it works - **Manual PCA pipeline.** Mean centering → covariance matrix → eigenvalues/eigenvectors → projection — every step explicit in NumPy. - **Real-world messy data.** 4,485 missing values, non-numeric columns dropped — handles actual CSV chaos, not toy datasets. - **Variance explained analysis.** Scree plot and cumulative variance — justifies component count with math, not guesswork. - **African health context.** Malaria incidence, water access, sanitation — interpretable loadings on real policy-relevant features. - **Formative submission.** ALU checker-compatible notebook with documented methodology section. ALU coursework · Christian Tonny