Final Projects From My Spring MOOCs

Spring has come and gone, and with it another wave of MOOC experiences.   This time, I also had the chance to submit a number of final project for peer assessments.  I am still skeptical about the reliability of peer assessed assignments, and I even attempted to use my teammates from Leading Strategic Innovation in Organizations to see the variability in our scores when we submitted the same assignments.

For example, we all submitted the same video for our final project: Innovations Final (v0). However, I was the only one of three team members who received the maximum possible score for the submission, as well as positive feedback:

Since the grades are relatively arbitrary and significantly based on luck, I won’t share my results, suffice it to say I received “passing marks” on the 4 peer assessed final projects I did submit:

Overall, all four of these classes were well worth the time spent. This Summer, I’ll hopefully add a few more projects to my library, including hopefully a Bitcoin Selfstarter if I can make it through all of Startup Engineering. Fingers crossed.



Sharing Your Results – Is It Cheating?

As of 2013, whenever you have a question, you Google it.

Likewise when you have a question about a MOOC you may be taking, it is often just as useful to Google it as it is to check the discussion forums. However, many people like to post their answers, and specifically, their results from MOOC projects for the public to see. And by doing so, the posters make it easier for new students to find the answers just by searching.

Therefore, if it’s not the first time the class is being hosted though, chances are pretty good that a student can find an exact post or file with the answers. And I’ve now found this to be true with a number of the classes I’ve taken; just search “coursera” or “edx” on GitHub. Specifically, I stumbled upon the following answers while I was seeking help on the internet (SPOILER ALERT: Links lead to answers!):

  • Stanford’s Introduction to Databases – Programming solutions can be found by searching the question on Stack Overflow
  • 10gen’s MongoDB for DBAs – The entire final is available on someone else’s blog (although not completely in English)
  • Johns Hopkins Computing for Data Analysis on Coursera – Projects with details of how analysis can be done are everywhere on GitHub

gitI am not helping the situation either, as one of my next posts will be a collection of final projects from the Winter/Spring “semester” I recently finished. However, I just can’t resist the temptation of putting my results out on the internet to be roasted by all those who roam the web. Although, I still haven’t decided whether it is morally right or wrong to share the information knowing that it may be used inappropriately.

Regardless, hopefully no one will plagiarize my work, although not much of it is worth plagiarizing anyways. But, I also post it in hopes that it might help some people avoid the same stumbling blocks I found, or maybe give them direction when they reach a mental impasse.

I’m sure there isn’t a “catch-all” solution, but I hope someone can answer me this, how can we collaborate through sharing and still prevent cheating?

The Coursera Signature Track Experience

April 2013: I decided to try out a Coursera Signature Track course. I (and others) question if a “verified certificate” is really worth it, but regardless I thought it would be interesting to examine just how Coursera “verifies” whether a student did indeed complete his or her own work.

Since UPenn’s Gamification course by Kevin Werbach was highly rated last Fall and available at a discounted price of $39, I decided to indulge.

Until I registered, I pondered how exactly Coursera would profile a student’s biometric signature for written assignments. My personal hypothesis was that they would track everything entered into text boxes for the written assignments, however this would make it impossible to write answers outside of the Coursera site, and impossible to grade the typical multiple choice quizzes. Apparently, that’s not the way Coursera’s system works at all.

Rather, you authenticate yourself via a text entry, a web photo of yourself, and a web photo of your photo ID (I used a driver’s license) when you first register. Then, a similar window will pop up after every quiz, assignment, or exam submission asking you to type the text and snap your own photo:

At the end of this Signature Track Gamification course, I’ll have definitely (maybe) mastered typing the sentence: I certify this submission as my own original work completed in accordance with the Coursera Honor Code.

Before the end, I’ll also have to try testing the system by asking others to type the sentence for me, or trying to use images of myself for the webcam photo. Hopefully my attempts to test the authentication system won’t disqualify me from ultimately receiving my verified certificate. If so, I hope I can at least capitalize on their lenient, but fair refund policy.

Regardless of how my own experiments turn out, this system is not without its flaws. It can easily be circumvented by merely giving your login credentials to another test-taker and having them “Save Answers”. Then the registered student can login again and just “Submit Answers” anytime before the deadline.

Despite my light bashing of the authentication system, I do believe this is a step in the right direction for Coursera and MOOCs in general. Verification of a student’s work is crucial, however, I hope a goal is to ultimately make “verification” universal and free of charge. At least Coursera is offering Financial Aid for those who need it for now. As for me, I’m more than willing to pay my small share to “play with the system” and fund Coursera’s biometric keystroke tracking R&D.

Fairness of MOOC Peer Assessments

A few weeks ago, I received results for my first Peer Assessment in the course Data Analysis on Coursera. I only had computer-graded quizzes in my previous courses, so I was interested to see the results. Personally, I and many others are fairly skeptical about the fairness and accuracy of these Peer Assessments. My reasons for skepticism:

  1. What prevents a student from sabotaging others’ grades?
  2. Conversely, what prevents a student from giving others full scores?
  3. What do your peers REALLY know about the subject?

I still don’t have any clear answers, but I do feel as though I myself have sabotaged others a bit. I received a grade of 82 out of 85, which also happens to be what I gave myself during the self-evaluation phase. But, I was required to grade four peers, who respectively got 58, 17, 57, and 53 from me. To try and justify that they were truly subpar submissions, I even graded an additional “optional” assignment in hopes that not all submissions were so bad. This last paper was much better; I gave it a 78.

More importantly though, what do most of the students (including myself) really know about the topic? For example, some of the papers used more advanced data analysis techniques, such as k-fold cross validation and boosting, but received worse grades. Probably because a majority of the students don’t understand what these techniques are, if they are better, and how they should be properly applied.

A slight confidence booster is that some professors do take notice of the discrepancies between normal grades and the Peer Assessments. In the course Developing Innovative Ideas for New Companies, the average grade for quizzes was 82% while the average for the Peer Assessment was 65%, thus the professor raised everyone’s grade by 17%. I’m not totally sold that this is the correct way to adjust, but it is comforting to know that professors are aware and involved in the process.

For those who are interested, my 4-page Data Analysis assignment, Lending Club Interest Rates are closely linked with FICO scores and Loan Length, is available, and it reveals nothing new or interesting to the world, but was good practice in building linear regressions. And, it meets the minimum requirements of the assignment.

Here are some other Data Analysis papers and their scores, for reference/comparison:


January-Loaded Spring Classes

So, after many attempts to slim down my January course load, I’ve finally narrowed myself down to 5 classes. And, hopefully I won’t add too many more as the season goes on.

Even now, I’m still trying to settle on one economics class.  I just can’t tell which one will spur my interests more.  The Economics for Scientists appears to offer much greater depth into the topic; however, the Microeconomics class was designed specifically for business students, so it might align more closely to my actual needs.

Coming up later in the year, I also have the following classes on the docket.  But who knows if my online Spring semester will include additional classes.

Unfortunately though, I do not expect to receive any kind of certificate or statement for the Business Strategy or Business Ethics classes.  Also, I need to assemble an entourage to take on the Strategic Innovation class with me, otherwise I will be ostracized and unable to join in the group activities.

Data Analysis and Financial Engineering will be good opportunities for me to really put my R skills to the test.  I also expect Economics for Scientists and Financial Engineering to be fairly challenging classes, so they might really change the rest of the semester’s lineup.  But I am also feeling somewhat compelled to take some classes that aren’t either quantitative or business-related.  I love a good challenge though, so we’ll see if the harder classes will slow me down.
2013 spring

Fall 2012 Semester Completed!

So I’ve finally finished my first “semester” trying to go to online class in a more formal manner. Before this Fall, I watched videos on my own, often through iTunes U, and doing homework assignments on my own where applicable, such as with the MIT OCW Scholar courses.

This Fall, I enrolled in the following classes on Coursera:

And unfortunately, I had to defer and drop out of the following edX class since I just didn’t have the time, but the 2 weeks I spent there was great, and I hope to try out their classes again in the future:

I must say I had a great experience with using Coursera.  For me, I am self-motivated and able to follow their weekly schedule of work, but I can see how it is hard for some.  It is just the right amount of structure for me to keep progressing, particularly via the weekly homework assignments and/or quizzes.  Some may argue that it is difficult to learn with such “disengaged” professors, but in these courses, I found the people on the forums to be more than helpful enough in answering questions.    Anyone who was enrolled in Operations could probably tell you that Collin somehow was able to help everyone despite being a student as well.

Computing for Data Analysis

There will always be an “honesty” issue as to whether or not a student actually did the work, so I’m still not totally convinced that these courses should ever qualify as legitimate college credit.  However, I think they are ideal for fulfilling pre-requisite type requirements. And, if you’re like me and are in it for the knowledge, then there’s no point in cheating because you’re actually enrolled to learn.  And in that case, it will change your world. No more tuition, no more late-night classes, no more summer school, and no more community college.