MITx "Machine Learning with Python" course additional to MLS?

Hi all,

I will finish the MLS during this week and I am asking myself how to continue after the course.

Background: I have a M. Sc. in Computer Science and am working since 20 years in the field of software development and architecture. However, I wanted to turn to a new direction, and so I decided to spend some time for education in data science. I finished the “IBM data science professional certification” and a Statistics course of Stanford, and soon the new MLS of Andrew Ng.

I also decided to start the " MicroMasters Program in Statistics and Data Science" of MITx and did the course “Probability - The Science of Uncertainty and Data” by Prof. Tsitsiklis. This course has proven to be “real university level”, which means it was quite hard work to pass the homework and exams and therefore I also have the feeling that I gained deep insights in probability theory.

Compared to this course, the above mentioned courses, despite presenting lots of material, were quite beginner friendly. I learned a lot in these other courses, but to pass the exams was quite easy. Now I am asking myself if I should enroll the next course of the MITx Micromasters program, which would be “Machine Learning with Python” (https://www.edx.org/course/machine-learning-with-python-from-linear-models-to).

I expect this course to cover - more or less - the same model types as the MLS. However, I expect the homework and exams to be much more work. Also, I would expect the depth of the course more deep with respect to mathematics and implementation details and maybe even conceptual details. Unfortunately a detailed course syllabus or even the course videos are not available as a preview.

So here is my question: Maybe, some of you have more knowledge about this MITx course and so you can help me with my decision: Are my assumptions correct that the MITx course “Machine Learning with Python” will provide me with even deeper insights and experience in the field of Machine Learning, which will prepare me better for a data science job later? Will it be worth the money and the time additional to the MLS and the IBM certification?

(I do not know if I hit the correct forum/group/category for this question, so pls help me to find the correct place to ask this if I was wrong.)

Thanks in advance for your perspective and any help!

Best regards
Matthias

1 Like

HI @Matthias_Kleine

You had done a great job but my advice is to start with more powerful algorithms which is deep learning and go deeper and deeper inside it as many the machine learning algorithms now is built in models in library but the future of the ai is deep learning and take start in a field that you find attractive to you in deep learning, and start delving into it more and more, such as computer vision or image processing , speech recognition or any other field that beside to make more and more projects in machine learning in the practical you will learn an advanced things and model …also I thinks there isn’t an course which will learn all things about the topic but he guide us to the right way and we continued after that …that is my advice you have freedom to take any decision. but I didn’t know about MITx Micromasters program

Cheers!
Abdelrahman

Hi @Matthias_Kleine, I am glad to hear that you are willing to devote the time and effort necessary to complete such a demanding courses.

Despite the fact that I did not take the Machine Learning with Python: from Linear Models to Deep Learning by MIT course, from reading the syllabus, I understand it to be equivalent to taking both Machine Learning and Deep Learning Specializations.

Therefore, I recommend that you complete Andrew’s Deep Learning Specialization first. Compared to the MLS, which you have already completed, this specialization will take much more time. I can personally attest that the DLS is an excellent program, which develops theoretical concepts and puts them into practice in a very effective and precise manner. Additionally, it would be a shame not to pursue the DLS after completing the MSL, since these two specializations are highly complementary and provides a great understanding of the subject.

I hope you find this answer useful.

Cheers,

Hello @Matthias_Kleine,

I have circled the topics/concepts that are not included in the MLS.

Sometimes I find insights to be knowledge that meet my needs at the time, and with exposure to more, I hope you might find what you are looking for :wink:

Happy new year, and cheers,
Raymond

PS: I am moving your thread to the General Discussion category for a wider audience.

Hi @rmwkwok ,

thank you for this great “diff” of the course contents. It really looks like the MITx team covers quite some issues that are not covered in the MLS.

I would like to add the following video review and comparison of the MITx machine learning course with the MLS/DLS courses from Andrew, given from some “childish” voice, which is probably artificially distorted. But the reviewer seems to have taken the whole MITx program and therefore has some insights that give further information about the program.

My preliminary conclusion is that the MITx course might indeed offer some additional content and also some deeper coverage of some issues of the MLS. It requires significantly more problem solving. And it also goes far deeper into the mathematical backgrounds needed to implement some algorithms (i. e. calculus and linear algebra).

However, it also needs much more time investment (3+ months, 15-20 hours per week) and is not self-paced.

I hope that the new Deeplearning.ai " Mathematics for Machine Learning and Data Science Specialization" will be published soon, so that I can review the contents of this new specialization. It might cover some of the mathematical backgrounds that I would really like to refresh and deepen. If this is the case, I will probably go the path MLS → DLS → MLOps → Math Specialization (not necessarily in this order) instead of following the MicroMasters program of MITx.

Best regards
Matthias

I think your plan is very reasonable. We are all looking forward to the release of the new specialization.

RNN and NLP are also covered in the DLS. Of the other topics that I circled, I think “EM algorithm” and “Generative model” are the most interesting ones!

The MLS covers K-means which is an example of the EM algorithm, but of course EM algorithm is a more general concept that applies in many other problems. It is simple and useful.

I don’t know how the MITx course is going to approach “Generative model”, but either presenting some statistical modeling or GAN is worth some time.

“separability” is a pretty interesting concept to have seen at least once.

Raymond

I agree with @rmwkwok: your plan sounds reasonable.

@Matthias_Kleine - FYI: this thread could be relevant for you, too:

Good luck and all the best for your learning journey!

Best
Christian