0
$\begingroup$

I've studied and understood well quite a bit of theoretical concepts of Machine Learning (except Deep Learning): e.g. the mathematics behind several classification and clustering algorithms, SGD and related optimization stuff, regularization, cross-validation, class imbalance and how to handle it. I've basic notions in Python and previously wrote simple codes in Python as well. I find the math behind ML easy, having an advanced degree in pure math, and I thoroughly enjoyed reading from books like Bishop's Pattern Recognition and Machine Learning.

However, since I'm applying for several data scientist positions, I'd like to practice building models and get my hand dirty to get a feel for industrial problems. Courses like Udemy doesn't seem to cover this, as they're mostly lectures with basic theory. On the other hand, Kaggle seems bit daunting for a total beginner. Hence, I'm looking for simple, previously solved, exercises like "Build a random forest and test with this and that metric", ideally something that can be solved within hours, not days (I hear Kaggle problems take time). Since I'm completely alone in this effort, ideally I'd like an exercise platform that contains the solutions or hints. I'd appreciate if anyone could suggest me some platforms, keeping in mind that I'm an absoulte beginner in practical ML.

$\endgroup$
  • 2
    $\begingroup$ Perhaps not exactly what you mean, but fast.ai has a very learning by doing problems (often kaggle ones) oriented machine learning (as well as deep learning course - especially the deep learning stuff has an extremely good reputation). $\endgroup$ – Björn Jan 5 at 14:21
  • 1
    $\begingroup$ I'd suggest finding some data science course materials from universities on the web that have solved homework problems. I did the following search and that led to some promising results: +"machine learning" +homework site:.edu University courses typically have homework assignments that must be turned around and turned in within a week, so this might be the middle of the road kind of thing you're looking for. $\endgroup$ – StatsStudent Jan 5 at 14:38
0
$\begingroup$

There can be three alternatives

1) Starting with a toy datasets like "Habermann's survival" on Kaggle and gradually attempting more complex problems

2) If you are more comfortable with a book and your focus is how to use Python and related frameworks and libraries then you can try "Practical Machine Learning With Python" by Dipanjan Sarkar, Raghav Bali, and Tushar Sharma. This book will help you to understand feature engineering and what are the possible algorithms for specific kind of data or situations. OR another similar book is "Introduction to Machine Learning with Python" by Andreas C. Müller and Sarah Guido

3) "An Introduction to statistical learning with application in R" by Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani - Code problems in this books are in "R" but they can be easily implemented in Python.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.