Quora Feeds
Active Member
Paul DeVos
None of them are the 'magic' bullet.
The TL;DR version:
There's no magic bullet. Spend 1-2 days on each of them. Find out what sticks for you in terms of LEARNING and INTEREST. If you want an ROI in terms of a credential, Anthony Lee's answer is also what I would say. Online Master's in CS from Georgia Tech with a ML specialization. Note, nine of the courses from the Georgia Techon Online Masters Degree (sans some materials) are on Udacity. So my advice, start with Udacity - if you really like those courses perhaps then go after the MS at GT. Otherwise, if you like Python I'd hit up the Scipy conferences (2013, 2014, 2015).
The Long Version
They're relatively unknown in terms of ROI as far as a "credential" goes. Sure, there were some pushes by the JHU team that helped get people who finished their MOOC specialization to get jobs, but the same can also be said of the Udacity Nanodegrees where Facebook, Google, and other big companies were involved with getting the courses up and running. I'm sure the hires in both cases were genuine. The question is, you can't just create a semi "pre-med" like equivalent to get a job at those companies or even companies in general. So there is no magic bullet - and that's no different than for a college degree. Sure, Ivy League credentials will more likely get you a 'shot' at a job [and at a higher starting salary] than a state college bachelors of the same major, but it's not a for sure case.
My Background - The lens in which I look through at these courses
Curriculum Writing Background
I have authored online courses in Biology, Chemistry, Physical Science, Algebra I, Algebra II, & Geometry for a top 50 high school in the state of Texas. I have also taught AP Calculus, Pre-AP Pre-Calculus and Physics. I've wrote all my own project based curriculum for those courses. I also helped author STEM education for K-12 educators for the State of Minnesota. I was also a part of GLPA (Great Lakes Planetarium Association) where I wrote materials in Astronomy for four years. I could go on...but you get the idea.
MOOCs Experience
I have been taking MOOCs since 2012. I've taken primarily CS, Statistics, Programming, and Machine Learning courses from Coursera, EdX, Stanford Online Learning, Udemy, and Udacity. And before that I was an avid user of other online materials to learn SQL and C# such as YouTube and Pluralsight. It's how I moved from the aforementioned teacher and get a job as a Data Analyst AND how I progressed from Data Analyst to Database Engineer to Data Engineer to Data Scientist in 4 years time. Franklin American Mortgage Company to Match.com to Google to IBM Watson.
I will note I don't typically finish courses. I'm self-taught in the CS/tech fields. Even in taking MOOCs I'm after specific things in learning vs 'whole course consumption' and will quickly move to another course if the scaffolding of learning is lacking. Time is the most valuable resource. Knowledge and learning are 2nd. I think a big missing part in assessing MOOCs is there seems to be such a focus on course completion and unfortunately on it's brick-and-morter brother, credentials. As a former teacher and as a person who now interviews potential candidates - I want to see people take their learning and apply it to projects on their own. I want to see the 'inspiration' move from beyond the course to the real world.
So onto the specializations!
Johns Hopkins Data Science Specialiazation
Ooof. The first course, Data Science Toolbox, is a joke of a class. They could have just sent you on a Google search to set up Git and R Studio. They should not take money for that class. R Programming is another dud IMO. The programming assignments have some rigor but it's low level examples and then WHAM - do fairly difficult level algorithm level coding for a beginner. I sampled the next three courses in sequence and honestly not worth the time I spent. There's some improvement, but not much.
If you want more detail, here' some reviews of those courses:
I would not recommend this course at all for Data Science or R. Not to any level learner; not beginner, intermediate, and certainly not an advanced learner.
If you want a taste of Data Science in R I would recommend two courses.
There is no comparison in these courses to the aforementioned R courses. None.
The Big Data Specialization by UC-San Diego
I took, wait, I endured the first two courses of this sequence. It's the highest level imaginable they could be taught at, except they weren't taught. It was read off the Cloudera website and other sources by the instructors. Stuff I would expect a person to have done BEFORE they signed up for these classes OR put as 'supplementary' reading materials you should read before taking this class. So this is also poor learning scaffolding, just as Johns Hopkins was, but in a different way.
Some reviews that I reflect a similar experience can be found here:
If you want to learn some Big Data technologies, go here:
Paul DeVos' answer to Is the Big Data Specialization from UC San Diego on Coursera worth taking compared to other online courses?
Illinois Data Mining Specialization
I only listened to a few lectures and found it wasn't something I could listen to. It also didn't use the tool I had focused on to master (Python). That's another topic, but pick ONE tool and do EVERYTHING or ALL YOU CAN in that tool until you master it. Then pick another if you so choose.
To my knowledge I don't believe all the courses have been released yet - so can't comment on the quality of these.
Udacity's Nanodegree programs.
I haven't enrolled in either of their Nanodegree programs. I have went through their Intro to Data Science and their Machine Learning courses. I liked both, but they will be the introduction level to your Data Science journey.
If you want my thoughts on the UW Machine Learning and those Udacity courses listed above please go here:
Paul DeVos' answer to Is it better to start with data science courses at Udacity (intro to data science) or the Machine learning course at Coursera?
Now, if you really want to learn Data Science and particularly in Python. I'd go to the Scipy websites for each of the past three years (2013, 2014, and 2015). There you will find the best of the best practitioners in Data Science for Python. If you can digest (like a Python snake) all of that and flesh it out - you'll be in HIGH demand. This will lead you to about 30-50 repositories on Github and actual EDAs (exploratory data analysis). You'll hear from the best of the best in the Python Data Science community. It's the Mecca of Data Science for Python.
Here's 2015:
Home | SciPy 2015 Conference
If you're not sure what language to choose I put some instructions to figure out how to tell your ROI based on available jobs [in your area(s)] for the technolog(ies) you desire to learn. You can find that answer here:
Paul DeVos' answer to Can I be a data scientist without learning Python?
Now there's no time like the present. Get after it.
See Questions On Quora
Continue reading...
None of them are the 'magic' bullet.
The TL;DR version:
There's no magic bullet. Spend 1-2 days on each of them. Find out what sticks for you in terms of LEARNING and INTEREST. If you want an ROI in terms of a credential, Anthony Lee's answer is also what I would say. Online Master's in CS from Georgia Tech with a ML specialization. Note, nine of the courses from the Georgia Techon Online Masters Degree (sans some materials) are on Udacity. So my advice, start with Udacity - if you really like those courses perhaps then go after the MS at GT. Otherwise, if you like Python I'd hit up the Scipy conferences (2013, 2014, 2015).
The Long Version
They're relatively unknown in terms of ROI as far as a "credential" goes. Sure, there were some pushes by the JHU team that helped get people who finished their MOOC specialization to get jobs, but the same can also be said of the Udacity Nanodegrees where Facebook, Google, and other big companies were involved with getting the courses up and running. I'm sure the hires in both cases were genuine. The question is, you can't just create a semi "pre-med" like equivalent to get a job at those companies or even companies in general. So there is no magic bullet - and that's no different than for a college degree. Sure, Ivy League credentials will more likely get you a 'shot' at a job [and at a higher starting salary] than a state college bachelors of the same major, but it's not a for sure case.
My Background - The lens in which I look through at these courses
Curriculum Writing Background
I have authored online courses in Biology, Chemistry, Physical Science, Algebra I, Algebra II, & Geometry for a top 50 high school in the state of Texas. I have also taught AP Calculus, Pre-AP Pre-Calculus and Physics. I've wrote all my own project based curriculum for those courses. I also helped author STEM education for K-12 educators for the State of Minnesota. I was also a part of GLPA (Great Lakes Planetarium Association) where I wrote materials in Astronomy for four years. I could go on...but you get the idea.
MOOCs Experience
I have been taking MOOCs since 2012. I've taken primarily CS, Statistics, Programming, and Machine Learning courses from Coursera, EdX, Stanford Online Learning, Udemy, and Udacity. And before that I was an avid user of other online materials to learn SQL and C# such as YouTube and Pluralsight. It's how I moved from the aforementioned teacher and get a job as a Data Analyst AND how I progressed from Data Analyst to Database Engineer to Data Engineer to Data Scientist in 4 years time. Franklin American Mortgage Company to Match.com to Google to IBM Watson.
I will note I don't typically finish courses. I'm self-taught in the CS/tech fields. Even in taking MOOCs I'm after specific things in learning vs 'whole course consumption' and will quickly move to another course if the scaffolding of learning is lacking. Time is the most valuable resource. Knowledge and learning are 2nd. I think a big missing part in assessing MOOCs is there seems to be such a focus on course completion and unfortunately on it's brick-and-morter brother, credentials. As a former teacher and as a person who now interviews potential candidates - I want to see people take their learning and apply it to projects on their own. I want to see the 'inspiration' move from beyond the course to the real world.
So onto the specializations!
Johns Hopkins Data Science Specialiazation
Ooof. The first course, Data Science Toolbox, is a joke of a class. They could have just sent you on a Google search to set up Git and R Studio. They should not take money for that class. R Programming is another dud IMO. The programming assignments have some rigor but it's low level examples and then WHAM - do fairly difficult level algorithm level coding for a beginner. I sampled the next three courses in sequence and honestly not worth the time I spent. There's some improvement, but not much.
If you want more detail, here' some reviews of those courses:
- Reviews for The Data Scientist’s Toolbox from Coursera | Class Central
- Reviews for R Programming from Coursera | Class Central
- Reviews for Statistical Inference from Coursera | Class Central
- Reviews for Developing Data Products from Coursera | Class Central
- Reviews for Regression Models from Coursera | Class Central
- Reviews for Getting and Cleaning Data from Coursera | Class Central
- Reviews for Practical Machine Learning from Coursera | Class Central
- Reviews for Exploratory Data Analysis from Coursera | Class Central
- Reviews for Reproducible Research from Coursera | Class Central
I would not recommend this course at all for Data Science or R. Not to any level learner; not beginner, intermediate, and certainly not an advanced learner.
If you want a taste of Data Science in R I would recommend two courses.
- The Analytics Edge (through MIT on EdX)
- Statistical Learning (through Stanford on Stanford Lagunita)
There is no comparison in these courses to the aforementioned R courses. None.
The Big Data Specialization by UC-San Diego
I took, wait, I endured the first two courses of this sequence. It's the highest level imaginable they could be taught at, except they weren't taught. It was read off the Cloudera website and other sources by the instructors. Stuff I would expect a person to have done BEFORE they signed up for these classes OR put as 'supplementary' reading materials you should read before taking this class. So this is also poor learning scaffolding, just as Johns Hopkins was, but in a different way.
Some reviews that I reflect a similar experience can be found here:
- Reviews for Introduction to Big Data from Coursera | Class Central
- Reviews for Introduction to Big Data Analytics from Coursera | Class Central
If you want to learn some Big Data technologies, go here:
Paul DeVos' answer to Is the Big Data Specialization from UC San Diego on Coursera worth taking compared to other online courses?
Illinois Data Mining Specialization
I only listened to a few lectures and found it wasn't something I could listen to. It also didn't use the tool I had focused on to master (Python). That's another topic, but pick ONE tool and do EVERYTHING or ALL YOU CAN in that tool until you master it. Then pick another if you so choose.
To my knowledge I don't believe all the courses have been released yet - so can't comment on the quality of these.
Udacity's Nanodegree programs.
I haven't enrolled in either of their Nanodegree programs. I have went through their Intro to Data Science and their Machine Learning courses. I liked both, but they will be the introduction level to your Data Science journey.
If you want my thoughts on the UW Machine Learning and those Udacity courses listed above please go here:
Paul DeVos' answer to Is it better to start with data science courses at Udacity (intro to data science) or the Machine learning course at Coursera?
Now, if you really want to learn Data Science and particularly in Python. I'd go to the Scipy websites for each of the past three years (2013, 2014, and 2015). There you will find the best of the best practitioners in Data Science for Python. If you can digest (like a Python snake) all of that and flesh it out - you'll be in HIGH demand. This will lead you to about 30-50 repositories on Github and actual EDAs (exploratory data analysis). You'll hear from the best of the best in the Python Data Science community. It's the Mecca of Data Science for Python.
Here's 2015:
Home | SciPy 2015 Conference
If you're not sure what language to choose I put some instructions to figure out how to tell your ROI based on available jobs [in your area(s)] for the technolog(ies) you desire to learn. You can find that answer here:
Paul DeVos' answer to Can I be a data scientist without learning Python?
Now there's no time like the present. Get after it.
See Questions On Quora
Continue reading...
