What is your review of Coursera Data Science Specialization Track?

[Quora Feeds]

Gaurav Sharma


★★★

I have completed all 9 courses and I liked most of them. I think they are the starting point in learning Data science. They tried to cover as much as they can but didn't cover deeply.

I never worked in R and R Programming course did give me basic idea of using R. Even though they didn't cover much about R but doing quizzes and assignments I learnt lot of stuff about R and then there was course on Getting and Cleaning Data and covered R deeply. Next course Exploratory Data Analysis also teaches you to use R for creating different charts/graphs but again you will have to learn by yourself for more deeper understanding.

Only two courses I liked least were Statistical Inference and Regression Models, may be they didn't provide basic foundations. It was important for me because I hadn't had in touch with Probability theories etc since long time.

These two courses gave me hard time because I had to go multiple websites to learn basic concepts. I would suggest to walk through the videos on Khan Academy if you are out of touch before taking above two courses.


Practical Machine Learning will not teach you the Machine Learning algorithms from scratch but just teaches you to how to use different R packages to perform Modelling and Machine Learning tasks on your data.

Developing Data Products, Reproducible research and Data scientist toolbox are easy ones. They are not most important ones but after these courses you will know different packages/tools/websites to effectively present your data science work.

See Questions On Quora

Continue reading...

 

[Quora Feeds]

Anonymous


★★

I am a software developer with no experience in R or in data science. I don't need these courses for my work but I thought I'd take them just for the sake of learning something new.

I successfully completed the first two courses "R Programming" and "The Data Scientist’s Toolbox". Both with distinction (earned 100%).

The courses gave a general intro to the field and to R. IMHO they were not good enough. Specifically, the R course was very eclectic and shallow. Much of the material required for the exercises was briefly mentioned in the video lectures, but what really bothered me is that the lectures felt like they were describing a set of loosely related subject instead of building a body of knowledge from the ground up, which is my preferred way to learn a language.

After completing it, I still don't feel I have a good grip of the basic principles and philosophy of the language, which is what I expected from a language intro course.

I decided not to continue with the track for now, not because the field is not interesting but because the courses were not good enough.

See Questions On Quora

Continue reading...

 

[Quora Feeds]

Chris Saden


★★★

I completed 8/9 courses in Johns Hopkins Data Science Specialization and took them for free in their first offering. I actually took the 9th and final course more details below. I did not complete the capstone project since you need to earn a verified certificate in all 9 courses. Final note for context, I took three classes at a time, which I don't recommend for those new to Data Science, Statistics, Machine Learning, and R.

I agree with what's been said here, especially in Nishant Gupta's answer to Is it worth it to pay 9 * $49 for a data science specialization on Coursera?.

My background: Math and Chem majors in college. Self-taught programmer. I took a number of online Statistics classes and Andrew Ng's Machine Learning class prior to completing the Specialization.

Full disclosure: I currently work at Udacity.


OVERVIEW

The courses teach how to use R and address some high level aspects of doing data science.

I don't think these courses are "beginner friendly." From Nishant Gupta's post...

"If you are new to the field of data science or analytics, this may not be the first resource you should go for. A decent understanding of statistics, ML techniques would be helpful in deriving the most out of these courses."

The courses are NOT great for learning the details of Machine Learning algorithms or gaining a deep understanding of Statistics.

The first three courses in the series are particularly easy, and then the difficulty starts to increase quickly. If you have little to no programming experience, you will likely get stuck on small steps in the programming tasks (which aren't always covered or explained in the course materials). I searched through the forums and asked questions to get unstuck.

If you are looking to learn R for data analysis, then I think the first four courses in the Specialization can get you up and running.

BEST PARTS

I found Exploratory Data Analysis and Practical Machine Learning to be useful classes. I walked away with skills and could share the final project as a demonstration of those skills.

I really enjoyed the course on Reproducible Research as well. Great to think about how you can collaborate with others and share results.

Exploratory Data Analysis teaches how to explore data through visualizations. The course teaches how to use the base graphics package and lattice package in R. I prefer the ggplot2 package for making graphs since I find it easier to use.

Practical Machine Learning teaches how to run Machine Learning algorithms in R and how to evaluate them. Lots to learn on your own in the course. The final project was interesting.


WORST PARTS

Two classes, Statistical Inference and Regression Models, lacked quality instruction and assessments. The professor mostly read from slides with a few worked examples. Regression Models was slightly better. I was also hoping for more opportunities to practice with the mathematical concepts. There were maybe 5 questions a week in the Statistical Inference course.

For the Statistical Inference class, too much content was covered within four weeks. As one student aptly noted, the course should be called "A Review of Statistical Inference". It would be incredibly difficult to learn the material in that class without having taken other Probability and Statistics courses. I found myself relying on two of my college level statistics courses to fill in the gaps.

Finally, the Shiny app example in the Developing Data Products was not inspiring. I didn't do the final project because I didn't have any strong ideas for an app. I think this was also due my own lack of motivation so I don't mean to criticize the instruction.

BEFORE TAKING THE SPECIALIZATION


1. Get exposure to R.

I recommend these options to get some experience with R. All are free.

Try R
swirl: Learn R, in R.
Exploratory Data Analysis Using R (I created this class with Facebook's Data Science team)


2. Build up your statistics background.

Patrick Conway's Statisics One [Princeton] (teaches how to do statistical tests and calculations in R)

Mine Çetinkaya-Rundel's Data Analysis and Statistical Inference (or DASI) [Duke]

The two courses mentioned above are rigorous and have high quality instruction and assessment.

Udacity also offers two courses in statistics. If you have never taken Statistics or need a refresher, the two courses below are great places to start.

Intro to Descriptive Statistics: Mathematics for Understanding Data
Inferential Statistics: Learn Statistical Analysis


3. Take a Linear Algebra course. Watch Gilbert Strang's lectures on MIT OCW and solve related problems on the OCW site.

Linear Algebra helps you understand the Regression Models course.


4. (Optional) Take Andrew Ng's Machine Learning course.
Knowledge of linear algebra and calculus will help you understand the mathematics behind the algorithms and how adjusting parameters affects the algorithms at run-time and their results.

See Questions On Quora

Continue reading...