So, I have just concluded the lectures of the MIT open course Probabilistic Systems Analysis and Applied Probability which has 25 lectures splitted in 4 main topics: Probability Models, General Random Variables, Random Process and Inferences.

My goal in the post, is not to tell about the concepts taught there. But to talk about my learning experience in this process. I think it was very challenging for me who just had the free nights to do so and only by myself. For sure, it was harder than a MOOC from Coursera, these certainly are more proper for a remote education. I knew that beforehand, but at the same time I wanted to take this shot to get a deeper experience.

At first, my main goal about this course was to learn the basic concepts of probability models as Random Variables, Probability Distribution, Marginal Probability, Joint Probability, Bayes Theorem. Since I have started my data scientist journey 4 years ago, eventually I was confronted with these concepts, and eventually I would google it and eventually it would take me to some Stackexchange site. There, usually I could get a grasp of the subject. However at the end of day I always felt a gap in my understanding about it.

I believe in long term learning process, and the basis of this is mastering the basic skills. So I took my shot.

Oh boy!, it took too long to finish it, almost one year following the course contents. Since the beginning I had in mind that I would follow this methodology: first read the chapter content of the class, watch the lecture, and do the exercises. In this order. And so I did it.

I like reading the material on books first and just later watch the class, I think this really improves the class experience. Afterwards, the exercises…. hummm… that’s the hardest part of the process, and surely the part that you’re most active.

At the beginning, I really didn’t escape that, because I knew the methodology would be the difference of how long I would take to forget the lectures. But the exercises really took too much time. Some were very instructives, others too much challenging (hard assimilation and heavy math). After a while, I had to ponderate the time that I put in these, because of my short available time.

Each lecture process (reading, watching, practicing) was taking me one week with 10 hours of effort. So in the long run, it would be IDEALLY 25 weeks with 250 hours of effort to finish the course. Oh sweet illusion, after a while, I faced a big problem: It is hard to take an exclusive routine for so long. Mainly, because I had other demands, including other topics that I had to learn in the meanwhile to solve my job tasks, or even personal demands.

Thus, in the middle of my journey I was struggling to dedicate my extra time only for that. And sometimes I was also taking too much time in the exercises. Suddenly my ideal one lecture a week was becoming two weeks. So after some struggles, I decided to be more pragmatic. I couldn’t be just wandering about a subject. I should turn it in a useful manner to my career.

And here I regard other main points that I added to a new approach. They are:

1. Get the grasp of the concept, a basic idea of what it is and what it isn’t;
2. What other concepts it is linked to;
3. Where you can search for it if you need an enlightment.

Following that, I fast paced the course. I also knew if it lasted even longer, it would be demotivating. So, after some more months, I finally finished the classes. Afterwards I better understood that: “It takes work to wander well”.

My main lesson from the course was how to learn really by myself. I have taken other MOOCs, specially on Coursera platform, but taking this MIT OpenCourseWare was a different deal. The Coursera material is usually shorter and more helpful for the extratime studying.

I think after some time, after college, after enrolling in the data science big area: Learning must be an ad hoc task.

About the Course subjects, I will not say details. However, I can say that the classes are very well taught and enlightening most of the time. The book used in the lectures is from the same professor of the class, John Tsitsiklis. In the end, it accomplishes very well the mission to introduce the student to the Probability world.

For now, I have that good feeling of having accomplished a very important homework in my data science journey. This year, I hope to go more practical. I aim a Deep Learning course, probably the fast.ai one, which offers a top-down approach. It’s time to cross to the other side.