I’m a former computational astrophysicist. I did that for 15-20ish years over 3 masters degrees and a bit beyond. I’m disabled by seizures and migraines right now, but hoping to expand my machine learning knowledge so that someday I can have a career again and so that I can do some data or science oriented computational project now. That’s why I’m in the Coursera course. I’ve got experience in C++, FORTRAN, MATLAB, C, and Python. I also have done a little in Java. I can’t say that I’m very good at it, and it was a long time ago, but I worked in HBase and parallel programming a little. Nothing useful. Since then, I’ve learned some SQL and Pandas, but at a beginner level. I have to admit that I really miss emacs and that Jupyter is driving me crazy. But I’m very excited to learn Machine Learning!!!
I’ve already tried a few other online programs, and have always had to stop due to software issues, whether it’s the cost of big data, or issues with interpreters that do not seem to run code on the website, but rather use AI to check the format. I was hoping to run code and debug using error messages and numerical judgment.
However, cost is definitely a big issue for me. So more recently I have simply been reading about machine learning and trying on my own. I have implemented a perceptron and Adeline from scratch and also used scikit-learn for those on the Iris data set and a small one I found on Kaggle. But I want to do more, and guidance and resources would help a lot. The books a slog, even if I am successfully learning, and a course would be motivating and provide feedback.
Re: Costs. That’s always a concern. Good courses tend to not be free, due to the cost of developing and supporting them, and providing a controlled learning environment.
The “UCI Machine Learning Repository” has lots of free sets of labeled data for you to experiment with.
There are a lot of free tutorial courses online and on a popular video sharing site, but they’re of highly varying quality.
Personally, you might look at the video channel for “3 blue 1 brown”. His material is very good on a number of topics.
I think KNeighborsClassifier, some of the other models, multiple layer neural networks(?), model selection, and tensorflow will be helpful to me. But thank you, that’s a nice thing to say.
I think I can manage the cost of the Coursera course. I’m just hoping there’s no big data costs. I was enrolled in another Coursera course previously that used Google’s big data service and I accidentally left or used up sandbox mode and had to drop the course. I also was enrolled in IBM’s course and the same thing happened. I also somehow left AWS’s free trial a long time ago. So I’m worried about similar issues, and I definitely can’t spend the thousands of dollars that can easily accumulate. $50/month is probably within my budget, though definitely not trivial given my income. $1000 is just completely out of the question.
Both of the sets of core courses (Machine Learning Specialization and Deep Learning Specialization) run entirely on Coursera’s platform, and there are no additional costs for data or processing access beyond the course fee.
But thank you, that’s a nice thing to say.