First learn the basics from books / manuals, especially the parameters which need customization. Second take some simple dataset e.g. Fisher-Iris dataset, and using some arbitrary test items configure the network to classify the test data. Third modify the parameters to observe what combinations produce the best results.
First of all, you need to learn about the basics. I would not recommend jumping to coding while you have no idea about how a simple NN such as MLP works.
I don't like coding (I'm not concentrating on that), but I've learned how to apply NN. Unfortunately, the books I've read are in Polish. However, my co-author asked me to describe the use of NN in our article DOI: 10.3390/ma12091396
My first software I've used was Excel file created by Augshuman Saha :-)
I'm attaching this file, as he stated it is free for studying purposes. Than my university provides Statistica software with NN module - there NN can be used without coding. For more advanced purposes I ask co-authors to code NN applications.