01
Nearest neighbors
k-NN classification, radius neighborhoods and feature scaling
Engineering portfolio · Bioinformatics foundations
Machine Learning.
From-scratch algorithm implementations and computational experiments exploring how machine learning works — from distance-based classifiers and decision trees to neural networks and combinatorial optimization.
29Python & Jupyter files
9Algorithm areas
PythonNumPy · SciPy · scikit-learn · Keras
Code archive
Algorithms & experiments
Each area links directly to the original source implementation. The emphasis is on learning the mechanics of models through code, evaluation and visualization.
02
Clustering
K-means, agglomerative clustering, density-based clustering and Dunn index
03
Decision trees
CART-style splitting and ROC/AUC feature evaluation
04
Ensembles
Bagged decision trees and cross-validated parameter search
05
Perceptron
Polynomial features and non-linear classification boundaries
1 source files ↗
06
Neural networks
Mini-batch logistic regression, MNIST preprocessing, CNNs on digits and letters
07
SVM
Linear, polynomial and RBF kernels; comparison with Random Forest
08
Probabilistic ML & regression
Gaussian and text Naive Bayes; polynomial least squares, Ridge and Lasso
09
Optimization
Monte Carlo, random walk, hill climbing and genetic algorithms for TSP
Technical capabilities
Methods explored
Mathematical modeling, feature transformation, training loops, performance measures and data visualization.
Algorithm implementationSupervised classificationUnsupervised clusteringGradient descentROC / AUCCross-validationFeature scalingNeural networksModel regularizationHyperparameter searchGenetic algorithmsScientific Python
Reproducibility note. This is an archive of historical coursework and research experiments, not a production-deployed ML system. Original datasets are not included, some paths and dependencies are outdated, and some implementations need correctness review before reuse. No unverified benchmark claims are made.