10 · NEURAL NETWORKS · HEALTH DATA · CLASSIFICATION

Vertebral Column Disease Classification

Overview

Built and tested a multilayer neural network on 310 samples and compared learning rates, momentum, and hidden-layer sizes before evaluating it on a held-out test set.

Details

Key result85.48% — testing accuracy

Stack

PythonTensorFlowKerasScikit-learnNeural Networks

85.48%testing accuracy

310samples

6input features

Next projectFitness Application — Video Loading Optimization →

Classification behavior on training and held-out testing data.

Training and testing F1 scores for the normal and abnormal classes.

How different learning rates affected training accuracy and loss.