I have not a very strong mathematical nor statistical background. This fact has not prevented me to work in a data analysis department because of my programming skills. I know the basics of important subjects such as calculus and linear algebra and I have implemented statistical analysis like logistic regression, support vector machines and many others, but always under the supervision of an statistics expert. I would like to learn advanced math and statistics in order to have a better understanding of the mechanics of what I am doing and most important WHY decisions are taken. Since I have time constraints, I have followed some MOOC's in Machine learning and statistics. But after completing the courses I had a bitter taste. I wasn't able to understand scientific papers because most of the authors assume advanced knowledge of real analysis, numerical methods, etc.

Clarification: Since this question has been put on hold I will try to give more details about this specific problem. I work from Monday to Friday. So I can only study on weekends. In the city where I live, there are no universities offering advanced math courses on weekends. So, firstly I bought some e-books on the subject but quickly realized two things: My progress was very slow and it was hard to tie up different subjects in an comprehensive way. I thought I could do better following online courses so I followed some MOOC's whose results I explained before. I'm convinced that the solution implies using Internet resources but by now I've not found a complete and satisfactory solution.

  • 2
    I'm not sure I understand the question. Why did the MOOCs leave you with a bitter taste? For example, was it because you didn't find them effective in teaching what they claimed to cover, or because you discovered that what they covered was not enough to read research papers? May 16 '16 at 17:11
  • 1
    Those MOOC's were not enough to read research papers. Most of MOOC's are focused on the procedures and techniques than the mathematics behind. For instance, none of the MOOC's I followed have even mentioned the relation between measure theory and statistics or the foundations of loss functions of mathematical models. May 16 '16 at 17:17
  • 2
    If you have a university close by you can usually utilize their library even if you can't get checkout permissions. Start with basic Real analysis texts and work your way through the catalog. You might even check out course descriptions to see what the more advance courses are using topic wise and work through those. Ask your stats buddies what books they would suggest. Write yourself out a schedule of topics you would like to learn. May 16 '16 at 17:26
  • I would if I had such an opportunity. Unfortunately, my schedule does not allow me to visit public libraries. Nevertheless, I think your two last suggestions are very useful. May 16 '16 at 20:25