Best online classes to learn Data Science
Data Science for Supply Chain Forecast
I receive many questions over the best online classes (also known as MOOC – Massive Open Online Course) on data science. This is for sure both a new field and a subject with many classes available. So what is the best path to learn data science online?
1. Learn Python
Your journey to learn Data Science should start by learning simply how to code in Python. You will need this basic skill to clean & analyze your data and then to make predictions (models) based on it.
In order to learn Python, I would advise you these two classes:
This is the python introduction class from MIT. It’s really clear and has a lots of exercices. The certification is also cheap (50€ or so). This is the #1 MOOC you should take. Only issue: nothing about pandas or data science. But this is a class about how to code in Python, remember?
This class follows the first one and goes a little bit further in terms of programmation algorithms. Perfect to be sure you understand everything about python, programming and optimization.
I decided personally to learn python rather R. Python is a growing language and is currently more used than R. But for sure you can do data science in R.
2. Learn “Python for Data Science”
“Python for Data Science” might be a strange title of concept but let me explain. If you went through the trainings from part 1, you just learned how to code in Python. Now you should learn all the tricks and techniques to perform data science in python. This will require some specific tools & libraries. To learn data science on python online I can only advise too MOOCs. Actually I don’t think these are good, just the best ones I could find so far.
- EdX – ColumbiaX – Analytics in Python (available for free)
- EdX – San DiegoX – Python for Data Science (need to pay 350$)
These two classes are both part of MicroMasters. Unfortunately (as explained below) I was very disappointed by the second class of ColumbiaX. So I wouldn’t advise to follow the Columbia MicroMaster.
If you find better MOOC on Python for Data Science, just let me know.
3. Learn Machine Learning
Now we can go into the interesting part. To perform analysis with decision trees, KNN or K-means normally the classes from part 2 should be enough. You can also take some practice on datacamp (but I only advise their trainings on machine learning).
If you want to go for neural networks, you should start with the theory. Andrew Ng (the co creator of coursera) made some excellent trainings. I really advise his excellent specialization on deep learning on Coursera (the class is in Python!). But this won’t show you how to create easily neural networks in Python. To learn this you should check the datacamp training on deep learning.
MOOC I don’t advise
I’ve also tried many other MOOC that I don’t especially recommend. Or even don’t recommend at all.
- EdX – Microsoft – Introduction to Python for Data Science. I never saw any good class from Microsoft on EdX and this one is no exception.
- DataCamp to learn Python. This is a personal opinion but I never could achieve good results by working on DataCamp exercices on basic python/pandas. Nevertheless I found their track on machine learning very practical & quite useful.
- EdX – ColumbiaX – Data, Models and Decisions in Business Analytics. This class was a total deception, no exercices, only short videos, poor tests. This is actually one of the very few class I decided to drop. Avoid it.
- Holt Winters forecast with multiplicative seasonality (Python)Nicolas Vandeput2019-11-13T16:41:48+01:00