Neural Network Projects with Python

Further improvements

At this point, it is worth wondering if it is possible to further improve the performance of our model. How can we further improve the accuracy of our model and/or improve the false negative and false positive rate?

In general, any limitation in performance is usually due to the lack of strong features in the dataset, rather than the complexity of the neural network used. The Pima Indians Diabetes dataset only consists of eight features, and it can be argued that these features alone are insufficient to really predict the onset of diabetes.

One way to increase the number of features we provide to the model is via feature engineering. Feature engineering is the process of using one's domain knowledge of the problem to ...