Applied Neural Networks
Coursework: three neural-network models for tabular, image, and time-series data.
Three models
NYC taxi fares
Tabular · Regression
A dense network that estimates fares from passenger count, trip distance, pickup time, day of the week, and pickup and drop-off coordinates, evaluated with R‑squared and mean squared error.
Handwritten Chinese numerals
Image · Classification
A 15-class classifier for 64×64 handwritten numeral images, with dropout, and error analysis through a confusion matrix and misclassified examples.
Melbourne temperatures
Time series · Forecasting
Simple RNN and LSTM models trained on rolling 30-day sequences, each producing a recursive seven-day minimum-temperature forecast, compared by mean absolute error.
The notebooks, charts, and results are in the repository on GitHub.