Mallory Learns

adventures in self-studying and life-long learning

[mallorylearns]

What’s the Point of College?

We think sending more people to college will revitalize the economy. What if it just creates a lot of bookish baristas?

Our current higher education ecosystem is going to be dead in a couple of decades at most. The cost of education cannot continue to skyrocket like this without providing a solid ROI. Eventually people are going to realize that higher-level degrees are not worth nearly as much as they’ve been led to believe, and they’re going to stop enrolling. Tuition will drop, some lower-performing schools will shut down, and fewer students will pursue bachelor/master/PHD programs. Shorter 1-2 year specialized programs, and apprenticeships, will continue to gain popularity with students and with employers and hopefully we’ll eventually have a new equilibrium that is more sustainable. I don’t think that’s the best outcome - because I still subscribe to the liberal-arts-citizen-building view of education and wish that the educational system could be designed to provide that rather than just turn a profit - but it is probably the best that the US can hope for at this point.

[mallorylearns]

Uber Would Like to Buy Your Robotics Department

When the company wanted a team of roboticists, it raided a university lab to get them. Can high-tech academia survive today’s Silicon Valley talent binge?

The deeper impact of Uber’s hiring may be how it affects what ideas roboticists will pursue. After all, 20 years ago, if you wanted to push the frontiers of robotics, you had primarily abstract academic options to consider. But now that the field is booming, a faculty member or grad student with an ambitious idea has to ask questions: Where’s the best place to pursue my research — at a university, or in the corporate R.& D. lab at a place like Uber? What type of engineering work is so far out, so hard and so unsolved that it can only be done at a university?

[mallorylearns]

City Mayors examines the advantages of linear cities

I think it’s cool that people are thinking of really ~out there~ ideas but this one is just too unitarian/homogeneous for me. Also some of the criticisms of current cities are totally unjustified.

What do you think?

[mallorylearns]

Protected Intersections For Bicyclists: this is the kind of urban planning I’d like to see put into action!

[mallorylearns]

Course Review: Machine Learning - Stanford University

Date Completed: April 20, 2016 Grade Achieved: 96.5%

Platform: Coursera (link to course) Workload: 4-5 hrs/week Length: 11 weeks Professors: Andrew Ng Would I recommend it?: yes

This was by far the most challenging MOOC I have taken to date and I loved it! You’ll want some familiarity with Linear Algebra and some basic statistics before beginning. Ng has designed a really wonderful course, covering much of machine learning: linear regression, gradient descent, logistic regression, neural networks, support vector machines, unsupervised learning, anomaly detection..., with great examples and challenging practice exercises/assignments that really allow you to absorb the materials.

There is extensive discussion of how to design a good learning algorithm going in, how to choose the best algorithm, and how to evaluate your algorithm’s efficacy and make improvements for both quality and speed. This was probably some of the most valuable info in the course as Ng is an expert and is basically just handing down knowledge that would otherwise take years of practice to figure out on your own. 

I enjoyed learning to work with Octave (essentially the open-sourced version of MATLAB, the syntax is easy enough but I am still learning about more of the coolest higher level functions that are made possible/simple in Octave) and larger data sets and am looking forward to completing some projects on my own with my newfound skills. Going forward I’m hoping to learn even more about neural networks in particular, and hopefully apply my knowledge to some fun personal projects or on Kaggle.

[mallorylearns]

Course Review: Greening the Economy (Lessons from Scandinavia) - Lund University

Date Completed: November 2015 Grade Achieved: 86.5%

Platform: Coursera (link to course) Workload: 1-1.5 hrs/week Length: 5 weeks Would I recommend it?: for younger students

I probably know a lot more than the average person on this subject, but I’m definitely not an expert. For anyone who has already spent some time thinking/learning/talking about climate change, you’ll probably want to skip this course.

I do think it would be a good course for younger learners, maybe ages 13-17, especially as a good jumping off point for both MOOCs and learning about climate change. I like the idea of looking to what Scandinavian countries are doing right that is keeping them so environmentally friendly, and how can the rest of the world shift toward that lifestyle, but unfortunately the course just didn’t go in-depth enough for me.

I wish they had skipped the weeks on individual choices and business-level choices and just spent more time on city- and country-wide policies and actions. Anyone with an interest in the environment knows that individually it’s better to drive less, bike/take public transit more; eat less meat; turn off the lights and switch to efficient bulbs; etc. Just skip to the part where you discuss what Scandinavian cities/governments are doing to make these green choices the default. Unfortunately, while it did address them during the last two weeks, the course didn’t spend as much time on policies and effective government actions as I was hoping.

Final note on the set-up of the course: there are a ton of different professors lecturing in this course, most of whom have accents speaking in English (it is a Swedish university; they’re all easy to understand at normal speed, but sometimes playing the videos at 1.5-2x speed made it difficult - just so you know if you’re thinking about the time commitment). Having so many lecturers also made the course seem like a large group project, where everyone has been assigned to do one section and nobody actually collaborates on the project as a whole. I think the course would have benefited from having fewer professors who could have made the course more cohesive.

I have a lot of criticisms but overall it was not a bad course, just not what I was hoping for.

[mallorylearns]

Keys to Successfully Completing a MOOC

Massive Open Online Courses are attracting more learners than ever - yet the average rate of completion is only around 15%! A whopping 85% of people who sign up for a class fail to finish the course. To date, I have successfully completed more than 7 MOOCs, though I have also dropped close to the same number. What makes the difference between a course I will pass and a course I won’t even finish? Here are some tips that have worked for me:

Spend time at the beginning of the course looking through the syllabus and (generously) estimating how much time you’ll need to dedicate to watching lecture videos, taking quizzes, and working on assignments.

Be realistic: if you estimate that a course will require 10 hours every week, but you’re in the middle of your busiest season at work and will be working overtime most nights, maybe you should delay your MOOC until things have calmed down some at the office. If you’ve signed up for 5 MOOCs in a flurry of excitement, think about which one(s) you’re really going to have time to complete. It’s better to pass 1 course than to take several only to drop them all halfway through due to stress.

Write down every assignment and quiz deadline in your calendar or to-do list of choice.

Set up reminder alerts for every deadline so you don’t forget (if you’re using a handwritten list/calendar, consider setting up timed emails to yourself for each assignment)

Block off time in your weekly schedule to work on the class based on your estimates from step 1. Again, think realistically about what time of day/week you can be most productive. Maybe this means waking up earlier some mornings to study, or maybe it means blocking off Sunday afternoons.

Make sure that the time you’ve blocked off is at least one or two days before the official deadlines for quizzes and assignments - that way, if something comes up and you have to postpone your study time, or if a project takes longer than expected, your assignments can still be turned in on time.

Finally, remember what motivated you to sign up for the MOOC in the first place! Something drew you to that course; just tap into that motivation any time you’re feeling like procrastinating.

[mallorylearns]

Course Review: Cities are back in town: Urban Sociology for a Globalizing Urban World - Sciences Po

Date completed: DROPPED

Platform: Coursera (link to course) Workload: ~2-3 hrs/week Length: 8 weeks Professor: Patrick Le Galès

First thing to note: this is a very French course in style. Meaning a lot of rambling lecture (take a look at the official course description text for an idea of what I mean) with minimal testing and only one assignment that determines your grade. The style isn’t for everyone, so you’ll have to try it out for yourself if you think you might be interested.

I only made it about halfway through this course. I actually found the course pretty interesting, but I was taking multiple other MOOC courses concurrently and there just wasn’t enough time. Each week focused on a different model of city/urbanization and how you can use each model to understand contemporary metropolises (to an extent). There was a mix of video lectures and integrated reading assignments.

My number one complaint with this course is that even though it’s on the Coursera platform, all of the lecture videos are hosted on Vimeo and impossible to watch at faster speeds. I saw a couple of other folks complain about this in the class forums too; there wasn’t even an option to download and play through VLC or another video player with speed controls. Prof Le Gales speaks pretty slowly, so I would have been able to get through these videos much more efficiently at 1.5 or 2x speed. Honestly, I think that could have made the difference between not finishing and finishing this course.

If you’re interested in urbanization and urban planning, I think this course could be worthwhile. I think the last weeks of the course that I never got to seemed like the more interesting models to explore, too (smart cities for example).

Final note: I probably could have technically “passed” this course as the final assignment required writing an analysis based on any one of the models explored, so I could have just picked one from the weeks I had watched. However, I felt that would be cheating and I don’t have any reason to need a certificate or “proof” that I’ve passed - without the actual knowledge from the whole class, it’s relatively worthless anyway, isn’t it?

[mallorylearns]

Course Review: Business Metrics for Data-Driven Companies - Duke University

Date Completed: November 24, 2015 Grade Achieved: 97.3%

Platform: Coursera (link to course) Workload: ~2-3 hrs/week Length: 4 weeks Professors: Daniel Egger, Jana Schaich Borg Would I recommend it?: maybe

This course was a pretty basic overview of business metrics and data science career options. It certainly covered some important material, so if you’re not at all familiar with things like dynamic metrics/strategies, revenue, risk, profitability, economies of scale, and basic accounting/financial ratios, you might want to do a little bit of research as you go along on anything that’s unfamiliar. If you’re an expert on these things, you’ll probably just want to skip this course.

I was already very familiar with most of the things discussed in this course (from my undergrad business degree and my professional career), but as I intend to take all the courses in Duke/Coursera’s “Excel to MySQL: Analytic Techniques for Business” Specialization (of which this is the first of four), I figured it would be good to get a review of the information that they think I ought to know. For that purpose, it was fine - looking back I probably would’ve been fine to skip it, but since it didn’t take long to go through the lectures and assignments, I’m satisfied with the review.

Basically, I’d recommend it if you’re interested in the topic but have little prior knowledge, but I definitely would not recommend it as a “just for fun” MOOC.

Based on my experience with this course, I think the rest of the specialization will be fairly easy but as I want to get an introduction to databases/MySQL (the final course in the specialization) and better data viz techniques (the third course), neither of which I have any background in, I am going to continue on.

[mallorylearns]

Course Review: The Age of Sustainable Development - Columbia University

Date Completed: October 17, 2015 Grade Achieved: 91.1%

Platform: Coursera (link to course) Workload: 1.5-2 hrs/week Length: 14 weeks Professors: Jeffrey Sachs Would I recommend it?: Yes

Prior experience: I did have some basic background due to an interest in this area, but it was not necessary to have any background to understand the content

While the content for the entire course is immediately available and it is designed to be self-paced, I actually enjoyed going through this course just a little bit each week on the schedule Coursera laid out. Prof Sachs talks slowly, so I often played videos on 1.5 or 2x speed, so the overall time spent on this course was quite low. I did take detailed notes while watching which helped me absorb the information and allows me to go back and fact-check when thinking and talking about these issues now.

The course focused on the Millennium Development Goals (MDGs) and Sustainable Development Goals (SDGs) set by the UN and the practical ways we as a planet can achieve these goals. Prof Sachs clearly presented the information with lots of facts and an A+ use of graphs/maps/data visualization.

I’m sure that reading any of his books on the subject would be a great supplement (or possibly substitute to this course, if you choose the 500+ page The Age of Sustainable Development). For this particular course, I think I absorbed more material from watching the lecture videos than I would have just from reading. I did recently obtain Sachs’ book The Price of Civilization and am looking forward to reading more from him.

I found Sachs to be extremely intelligent/well-spoken/easy to listen to and the lectures full of new facts. This was not just surface-level overview and you can tell that the course has been very well designed, in terms of everything from syllabus to video quality. Sachs knows his stuff and does a great job transferring that knowledge to the virtual learner.

The only thing that’s keeping this from being as good as an in-person college course would have been is the lack of research projects and essays - but that just means it’s up to the MOOC student to do further research and/or writing to the extent that it interests them.

[mallorylearns]

Course Review: The Data Scientist’s Toolbox - John Hopkins University

Date Completed: May 11, 2015 Grade Achieved: 100%

Platform: Coursera (link to course) Workload: 1-4 hours total Length: 4 weeks in theory, 1 evening in practice Would I recommend it?: No

Prior experience: none

This “course” is a nothing more than a super basic guide to installing R and Github. The “conceptual” overview is laughable. This is essentially the 10 minutes “welcome to class” intro you get on the first day of school, not even the next ten minutes where they discuss the syllabus. It makes a little bit more sense when you realize that this class is part 1 out of a 10 course Data Science Specialization on Coursera, but after beginning the second course (Programming in R), I came to realize that the whole 10-course specialization probably doesn’t even cover a semester’s worth of introductory knowledge. Don’t waste your time with this one.

[mallorylearns]

My final project for Interactive Programming in Python: a game of asteroids (just a screenshot, since sharing my code would be a violation of the class Honor Code)

[mallorylearns]

Simple Blackjack game created for Interactive Programming in Python course (just a screenshot because sharing my code would be a violation of the Honor Code)

[mallorylearns]

“Memory” game created for Interactive Programming in Python course (just a screenshot because sharing my code would be a violation of the class Honor Code)

[mallorylearns]

Course Review: An Introduction to Interactive Programming in Python (Parts 1 and 2) - Rice University

Date Completed: October 19, 2015 (Part 1), December 30, 2015 (Part 2) Grade Achieved: 91.8% (Part 2), 74.5% (Part 1)

Platform: Coursera (link to course) Workload: ~5-8 hrs/week Length: 9 weeks (total for both Part 1 and Part 2) Professors: Joe Warren, Scott Rixner, John Greiner Would I recommend it?: Yes

Prior experience: I had gone through Codecademy’s Python path a few months earlier, and dabbled a bit in some other simple things in the meantime, but I was still was very much new to Python and programming in general. I think this little bit of extra background allowed me to coast through Part 1 without much effort at all.

You can tell that the profs in this course are very passionate about it and they’ve put in a TON of work to make this MOOC work really well. This course uses a web-based programming site called CodeSkulptor which was created by the professors. This makes it super easy to just begin coding without having to set up your own environment which is great, and it has very easy to use documentation so you can always figure out what you need to do for an assignment. I wonder if it would have been more beneficial for me to slog through a more difficult set-up process in order to have something I could use in a “real-life” setting. I’m also a bit worried that a lot of the libraries and functions we called to within CodeSkulptor aren’t the same as what you would use running Python normally, but I think the general knowledge/skillset will transfer over easily.

The material in this course is released week by week and you have to adhere to a strict schedule for turning in and peer-reviewing assignments - if you just want to go through the material as quickly as possible (perhaps if you already have some experience with programming), you can dig up an archived version of the course, but I found the weekly schedule to be really effective/pleasant for this course.

Each week in this course, you code an interactive game project - obviously getting progressively more difficult as it goes along. They provide the graphics, etc, so you can just focus on the coding and the concepts that are important to think about when creating a program. This format worked really well, and definitely keeps you feeling accomplished and motivated each week. While I didn’t originally have a huge interest in programming games specifically, I now feel that I have a much more solid understanding of how Python (and coding in general) works which I think will serve me well in the future. I feel quite confident with for/while loops, if statements, classes, lists, and other concepts that had evaded me before this course. Bonus: I’m just beginning a Machine Learning course and I’m already finding that my Python knowledge from this course is helping me understand how to use Octave for that course.

I’m also thinking about creating a simple game in the future, now that I see how easy it is! (For example, a simple game like Flappy Bird was probably super easy to code but is really fun to play - I’m not trying to create the next viral game but it could be a fun project for a weekend!)

In conclusion, this is one of the best MOOCs I have taken so far. I would recommend it for anyone who wants to get an intro to programming.

You can see some of my projects for this course here.

Notes: It took me longer than 9 weeks to complete this set, as I took about a month off between parts 1 and part 2. My grade in part 1 is also quite low as I turned in some assignments late and/or skipped the peer review sections - actually, this is also the case in part 2, I think almost all of the points I lost were from forgetting to do a peer review. I know it’s important to do peer reviews in order to keep MOOCs sustainable, but on the other hand I do not feel that the number grades always reflect the level of learning due to these kinds of markdowns.

[mallorylearns]

Course Review: Gamification - University of Pennsylvania

Date completed: April 15, 2014

Platform: Coursera (link to course) Workload: ~2-3 hrs/week Length: 5 weeks Professor: Kevin Werbach Would I recommend it?: 50/50

Prior experience: None, vaguely familiar with the concept of gamification

Taken during my final semester of college, this was my first introduction to the concept of MOOCs. Gamification was (is?) a hot topic at the time and I found the content quite interesting. The workload for this course was minimal, just a couple of hours a week for about five weeks. However, I probably could have learned just as much - and learned more quickly - by reading a solid book on gamification. The course was actually created around a book (For the Win: How Game Thinking Can Revolutionize Your Business) written by the professor, Kevin Werbach, and I would probably recommend just reading that or another gamification book rather than taking this online course. However this course did spark an interest in both MOOCs and gamification which have continued to this day, so for that I have to give it some credit.