Machine Learning Portfolio

A repository of deep-learning and machine-learning projects covering computer vision, time-series forecasting, and structured-data classification.

The problem

Models are only useful when their data, evaluation method, and limitations are explicit.

Source notebooks, datasets, experiment outputs, and project notes are available in the public repository.

Built and evaluated Simple CNN and VGG16 models for UTKFace age and gender prediction, including data augmentation and configuration experiments.
Created LSTM forecasting experiments for BHP, CBA, and WES stock data, comparing neuron counts and epochs with prediction graphs, RMSE, and R².
Developed an olive-oil classifier using preprocessing, PCA, Gaussian Mixture Model balancing, and Adam/SGD comparisons.