What was it like to take Computing for Data Analysis (CSE 6040) course in Fall 2017?

Selección_046.png

Computing for Data Analysis is a data science introductory course from Georgia Tech at edX, part of the MicroMaster in Analytics worth 9 credits accountable for the Online Master of Science in Analytics from the same university.

This course teaches you Python from the very beginning, in fact it was not a difficult course, but don’t be fooled by that. It was work intensive indeed. Let me share with you a progress graph that shows the quantity of assignments I had to do to finish the course.

Selección_047.png

If you are an edX student like me, you can see this graph clicking in the “Progress” tab of the courses you’ve taken. As you can see, there were 15 notebooks this Fall 2017, 16 if you count the Midterm 1-R. There were 2 Midterms and 1 Final Exam.

This is how grade was divided:

Selección_048

As shown, there were so many notebooks, but they only accounted for 50% of the grade, 3 exams were 50% of the grade.

Just as a side note, you can find me through my Linkedin profile and ask me any question related to OMSCS, OMSA or online education in general. Just send me an invitation and I’ll be glad to share my network with you.

I liked this course, because it taught me things that I think are very basic every day usage concepts that I hadn’t had the chance to polish on. I’ll share with you the list of topics:

  1. Python Essentials (there were some things I didn’t know of, like functional programming)
  2. Pairwise association mining
  3. Math prerequisites review (a lot of linear algebra and matrices stuff)
  4. Representing numbers
  5. Preprocessing unstructured text (always so necessary)
  6. Mining the web (useful indeed)
  7. Tidying data
  8. Visualizing data and results
  9. Relational data
  10. Intro to Numpy/Scipy for numerical computation (I had already gone through this, but a little brushing up wasn’t bad)
  11. Ranking relational objects
  12. Linear Regression (a little Machine Learning to make things more interesting)
  13. Classification (more ML)
  14. Clustering via k-means (still more ML)
  15. Compression via PCA (ML, ML, ML)
  16. Eigenfaces (personally I hate this topic, but very recurrent in Analytics courses indeed)

You can take a look at my course certificate if you want. That’s how it would look yours if you take the course or any other edX course.

Pros:

  1. I liked the course, it touched very foundational topics and my favorite language after php and javascript: python, what an important and fun language to learn.
  2. I liked how we used Jupyter notebooks, I hadn’t had the chance to work with them. Vocareum is a web jupyter notebook server I didn’t know, a real jewel.
  3. I liked the level of automation in this course, almost fully automated. We could instantly know what our grades were after every response we gave, all thanks to Vocareum.

Cons:

  1. From my point of view, there was relatively short time to address midterm 1, which was in Vocareum, there was a lot of reading to do to understand what to do on each question. I have nothing against to coding as part of an exam, even though it was the first time for me in this way of doing things, but time was an issue, at least for me.
  2. Some topics like “Representing numbers” were a little boring to me.
  3. I took this course as a verified student, there were other 2 types of students: audit students and Georgia Tech students that took the course as part of the Master of Science in Analytics. GT and verified student forums were put in Piazza (a very handy forum software used by other famous courses) in 2 different groups, with majority of students coming from GT, so verified students were kind of isolated from the majority of the perspectives on the course topics, which made learning not as fun as other online courses I’ve taken.

In sum, this course took a lot of time from my schedule in the 15 weeks it lasted, from 5 to 10 hours per week maybe. Not as much as other courses, but you already saw the number of assignments that were allocated. If I had the chance to take it again, I would do it, it was fun and it strengthened basic foundations for me as a data analyst. I felt I had to tune my python knowledge a bit, and this course did just that. Hope you like it as much as I did if you are about to take it. Best of luck! Don’t forget to connect with me in LinkedIn, visit my public profile. I’ll be glad to share my network with you. See ya.

What was it like to take Computing for Data Analysis (CSE 6040) course in Fall 2017?

What was it like to take Data & Visual Analytics course (CSE 6242) in Fall 2017?

dva

I wasn’t sure to take Data & Visual Analytics course this Fall, as there had been some negative comments in the OMSCS Google+ Group for Spring 2017, but everything went smoothly. You can take a peek to the DVA course website. There were 3 homeworks and 2 projects and just 1 final exam as you can see below. There were other 5 assignments called Activities that were worth 1% each, and they were completely optional as you can see if you sum everything up. Activities were not 10 times easier than homeworks (worth 10% each) let me tell you haha.

Selección_043.png

The exam was open notes/internet, it was still proctored using ProctorTrack as usual, but just because they really wanted to make sure that you didn’t get someone helping you out in person or through the internet (something like a forum or slack request for help) during the exam.

Letter grading rules were nothing special really, you got an A for 90 or above and a B for 80 or above.

Selección_044

I did 4 out of the 5 activities, that assured me 4 extra points, which I though may be useful as I remember someone telling me the final exam worth 30% could be a massacre if there was no curve applied, which is just what happened in previous semesters. I didn’t want to be so exposed to a B or C grade chance, as there was a different course teacher (Dr Joyner), and we didn’t know much about what to expect specially on final exam.

Just as a side note, you can find me through my Linkedin profile and ask me any question related to OMSCS, OMSA or online education in general. Just send me an invitation and I’ll be glad to share my network with you.

I spent hours on each “1% activity”, and I remember that after I finished each of them I got like “wow and this was only for a 1% of the grade”. But in the end I think it was worth it, I got an A as I wanted.

I think what made things a little easier is that I had already taken Machine Learning (CS 7641) and Machine Learning for Trading (CS 7646) courses before DVA, which helped me a lot, specially for the final part of the course where we saw Logistic and Linear Regression. Additionally, something I think helped me was that I was taking Computing for Data Analysis (CSE 6040), another Georgia Tech course at edX at the same time, which covered maybe a 30% of the material, but from a different perspective (using Python instead of R and from a computational angle).

Another observation would be that really all Data Analytics courses I’ve taken so far have touched Machine Learning concepts in some way, most of them include it by the end of the course material, and it really makes things more interesting and a little more difficult as well.

Something particularly stressful that happened to me while taking DVA was having 50% of the grade on the table by the time I took the final exam, but everything turned out just fine.

In the end, DVA wasn’t as tough as I expected, coming to a very unknown environment and with some hard comments on Spring 2017 course roll out. My expectations were low, but in the end I got a course that wasn’t as difficult or problematic as I thought and I had a lot of fun with data visualization and I got to practice R a little more, as I had already had an R course in Summer, which I recommend to you (GT’s ISYE 6501x Introduction to Analytics Modeling at edX).

PS: If I had the chance to take courses in a different order, I would’ve taken DVA and ML4T courses first, and ML afterwards, as ML was by far a tougher course. Just to make a comparison if you’ve already taken ML4T, I consider ML4T to be more work intensive than DVA, but not necessarily more difficult.

Well, guys this is all, have fun! Good luck picking your courses for next semester. Hopefully I’ll be taking 3 courses next Spring 2018: Graduate Algorithms, Big Data for Health Informatics and Human Computer Interaction. I’m a little frightened by the difficulty of GA and BD4H and taking them at the same time but I have no other option, wish me luck. See ya.

What was it like to take Data & Visual Analytics course (CSE 6242) in Fall 2017?

Google Colaboratory, a new tool to spread Machine Learning to the world

Google Colaboratory, a new tool, something like Github for Jupyter notebooks. It kind of looks like Vocareum.com to me. You can use TensorFlow natively, neat! It only supports Python 2.7 and Google Chrome desktop at the moment, but it sounds promising. The jupyter notebook files are saved in Google Drive. More information about it you can find at its FAQ: https://research.google.com/colaboratory/faq.html

Google Colaboratory, a new tool to spread Machine Learning to the world

How to improve your chances to get into OMSA (Online Master of Science in Analytics) from Georgia Tech

I haven’t applied to OMSA just yet, as I’m still finishing OMSCS (Online Master of Science in Computer Science). No one can be in 2 masters in GT at the same time, not my rule. This is from my experience as a GT grad student and as a verified student of GT’s MicroMaster in Analytics at edX.

First things first, nothing guarantees you’ll get into OMSA, but of course you can still build a strong application and improve your chances significantly.

Some stats from first OMSA cohort

Admission Committee will look at your profile and try to find some proof of a person who will finish the program (36 credits). For them to trust you in accomplishing that you need to demonstrate knowledge, experience, skills and more importantly a strong will, believe me you’ll need them all at some point.

Just as a side note, you can find me through my Linkedin profile and ask me any question related to OMSCS, OMSA or online education in general. Just send me an invitation and I’ll be glad to share my network with you.

If you don’t know or don’t remember some aspects of computer science or Statistics principles, I suggest you get into one or several courses, maybe some MOOCs in Coursera, Udacity or edX, as soon as you can. Mentioning them in you Statement of Purpose is a must.

GRE

From image shown below you may think GRE is not that important, as of 100 people who took the survey, most admitted students didn’t take GRE test. But at the same time, most people who didn’t get admitted didn’t take it either. I would suggest you take the GRE exam and try to ace it.

Image above was taken from here

GT’s MicroMasters in Analytics

Micromaster in Analytics from Georgia Tech at edX has 3 courses, all 3 from OMSA, they cost $500 each. If you’ve got a coupon, you can get each course for as low as $250. In comparison, an OMSA course costs $800 from what I know. There is a significant difference there. In that web page you’ll find the text shown below:

The Analytics MicroMasters program is comprised of three foundational courses from Georgia Tech’s full, interdisciplinary Online Master of Science in Analytics (OMS Analytics) degree available for less than $10,000. The MicroMasters program can serve as a helpful introduction to the degree experience.

Anyone with a Bachelor’s degree or equivalent may apply to the OMS Analytics program, but the successful completion of one or more of these MicroMasters courses can help your application stand out in the admissions process. If admitted, learners have the opportunity to apply up to 9 hours (3 courses) of the edX MicroMasters credential as advanced standing credit toward the total of 36 hours required for the degree.

At the moment of this writing, there is only 1 of 3 courses available to register, the other 2 were available at some point. You need to be a verified student and scan your government ID in order to be eligible to ask for credits for each course (3 credits each) once you’re admitted to OMSA. If at first you are not presented with the “verified student” option, don’t worry, it means it will be available later on. Just be patient and check once in a while before start date, and have your coupon and credit card ready to do the payment. If you take the course as an audit student you won’t be able to ask for credits afterwards.

I think it’s not necessary that I tell you that you need to ace these courses, try to take at least one, and mention it in your Statement of Purpose (SOP). More on SOP below. If you can only take 1 MOOC before or while you build your application, take 1 of these courses, it doesn’t matter if the application process finishes before the course does. Mention your edX account ID as part of your SOP so they can jump in and see how you’re doing.

micromaster.png

Statement of Purpose (SOP)

Maybe you would like to read this book to build your statement of purpose, it costs $3 in its kindle version at Amazon. I get nothing if you buy it, no money at all, buy another book or no book at all if you want to. You don’t have to read it all, but I encourage you to get one of this kind of books, it will make your SOP stand out from the rest. In my country there is not such thing as SOP to get into college, but for americans it seems to be a very important piece of information for admission decision. An SOP can make or break your whole application. If your grades as an undergraduate student weren’t the best you need to tell them why, and how you’ve changed since you got out of college, show some proof. Remember the number of applicants that actually get accepted from the video above, you want to be one of those guys.

Join the OMS Analytics Community

There is a Google Group called Georgia Tech OMS Analytics, I suggest you join that group, and ask for advice in any topic that hasn’t been mentioned in this post. There is an unofficial Slack group as well, but I think you need a GT email to be accepted by default, so you may need to be accepted by someone first. When you just start to get into something, any information is good information. Look for help.

Conclusion

You may get admitted without following any of the advice mentioned here, it is absolutely possible. But if you don’t want to wait for 6 months to apply once again, you better listen to advice from all people who’s made it through the admission process already. Not just me, I got into OMSCS, not OMSA, there is a lot more people who knows about this more than I do. Try to reach out to them and ask what they think.

Many other things can be added to this post I’m sure. If you find something, please let me know and I’ll add it. Thank you. Good luck with your OMSA application.

How to improve your chances to get into OMSA (Online Master of Science in Analytics) from Georgia Tech

¿Cuándo usar desarrollo iterativo?

2.jpg

“¿Cuándo usar desarrollo iterativo? Debes usar desarrollo iterativo solo en los proyectos que quieres que sean exitosos.” – Martin Fowler

Desde mi punto de vista, si estás a las puertas de iniciar un proyecto, y el equipo no sabe trabajar de manera ágil y peor aún, no tiene la cultura ágil, es mejor no usar ágil, porque puedes generar el efecto contrario al que esperabas. En lo que sí creo es que si el equipo supiera y quisiera trabajar en ágil, terminaría antes y daría mejores resultados que un proyecto tradicional, y de largo.
A continuación algunas FALSAS justificaciones que yo he escuchado para no usar ágil:
  • Agil no tiene calidad
  • En Agil no se planifica
  • Con Agil no puedo estimar fechas de entrega
  • Agil solo sirve en proyectos de desarrollo de software

Imagen de Mountain Goat Software. Traduccion de Juan Vizueta

 

¿Cuándo usar desarrollo iterativo?

Are you meeting your customer needs?

1f74f74ee3140632b63475a21fab9900

How many times have we seen people running to finish what they have been asked for? Only to find out not too long after finishing that customer barely used it or didn’t really use it at all. Did you know that is what happens to 64% of features in products? Then what should we focus on? I want to know what you think, leave me a comment below.

 

 

 

Are you meeting your customer needs?