Improving Keras classifier, dealing with dataset, getting wrong accuracy with Pandas, sklearn

Question

I want to build a classifier that takes into account multiple factors. I have been trying to follow tutorials and examples, but they all use data that is strictly numbers. T