Apply an Artificial Neural Network model on Wisconsin Breast Cancer Data to predict if tumour is Benign or Malignant.
Data set consists of 699 instances with 9 relevant attributes. Training and testing split is 80-20
For More Information on the Data: https://archive.ics.uci.edu/ml/datasets/breast+cancer+wisconsin+(original)
The Neural Network will contain 3 layers
1 input layer with a rectifier as activation function (9 nodes)
1 hidden layer with a rectifier as activation function (5 nodes)
1 output layer with a sigmoid as activation function (1 node for binary prediction)
The initial Testing Accuracy of the model was 0.963
However, I used 10 fold CrossValidation on the model for a more reliable testing accuracy.
It resulted in this set of accuracies: [0.98181818, 0.96363637, 0.96363635, 0.92727273, 0.94545454,0.94444444, 0.96296295, 1, 0.94444443, 0.98148148]
with a mean of 0.9615, which is close to the initial testing accuracy.
Mihai Groza