I have 52k rows of data including test and cross validation.
Features less than 10.
Which is better just logistic regression or neural network?
It depends on how complicated the problem you’re trying to solve is.
The choice doesn’t really depend on the size of the data set.
Yes, as Tom says, it’s not the amount of data you have, but how complex the “decision boundary” is between the “yes” and “no” examples. Of course if you have 10 features, then it’s a little hard to visualize (e.g. by graphing your data) what the decision boundary will look like in 10 dimensional space. Our brain just has a difficult time visualizing things in more than 3 dimensions. 
Logistic Regression is a simpler algorithm and less expensive to run and to train, but it can only create linear decision boundaries i.e. “hyperplanes” in your input space.
A Neural Network on the other hand is more expensive, but also much more powerful and can represent essentially arbitrarily complex decision boundaries at least from a theoretical point of view.
So the general approach would be to start by trying logistic regression on your dataset and see what level of accuracy it achieves. Maybe you get lucky and it is “good enough” for the problem you need to solve. But if it’s not, then you try neural networks, starting with relatively small ones (2 or 3 layers) and see what kind of results you get. There are a number of places in the courses here where this type of issue is discussed in detail. E.g. in DLS Course 1, Prof Ng starts with an image recognition problem and shows how to solve it with Logistic Regression and then how to get better and better solutions by using Neural Networks.