Why Do I Create Free Data Science and Machine Learning Educational Content - For Revenge
Why do I spend so much creating freely accessible educational content on data science? Many reasons. The most immediate are my professional development (teaching others is the best way to learn “for me”), to showcase my skills to potential employers, and because I just like it. But there is another one: revenge.
“Revenge? Seriously? That’s weird compadre…” Well, maybe. You see, during my educational experience, I was told that mathematics and “hard” sciences were probably beyond the scope of my learning skills and intelligence. Directly and indirectly. Even during my PhD studies, people have told me that taking class “X” is probably too hard for me. “Sure Pablo, but I bet other people were encouraging too” Absolutely right: I have had both encouraging and discouraging people in my life, telling me that “of course you can learn anything!” and “Maybe this is too hard for you”, respectively. So, what is the issue? After all, you can simply ignore discouraging people. Right?. This may or may not apply to you, but I was not the kind of child and teenager who could simply brush away discouraging and hurtful commentaries. When you grew up as a poor kid, in a poor country, with uneducated caregivers and unskilled educators, amid violence and neglect, learning to ignore negative feedback it is difficult. Also convincing yourself you can learn anything turns out to be hard. My point is: depending on your circumstances, discouraging and harmful commentaries about your skill and intelligence can have a “disproportionally” large negative effect in your learning outcomes and feelings of self-efficacy. “Well, maybe this is jus you!” True, I am claiming this based on my personal experience (and what I have observed in others), but there is a large corpus of research in psychology and educational sciences showing that it is not just me (I wont’ cite the papers here, I am kinda tired of writing everything as a scientific article). It took me many years of effort, luck, and small victories to convince myself I could indeed learn anything if I put the time and effort.
My first memory of learning math is my father’s frustration and anger because of my “inability” to understand how place value works in an abacus. I was so nervous and scared of my father than I just could not put my head to work. Those study sessions did not instill a lot of self-confidence in me. Teachers at school were not much better. Many operated under the absurd idea that they were some students for whom math came “naturally”, and the rest, well, the rest should try very hard and hope for the best. Society needs lawyers and historians too. Then high school happened. I almost failed my first and second year of high school because of math: I ended up the first semester, in both years, with failing grades in math. A failing grade in my high school meant having to repeat that year. But I did not. Failing students look bad for public school. This is what they did: they hand me over an “extra” final test and said that if I got at least a 4 (the minimum passing grade in the Chilean system), they will let me pass to the next year. I did, both times, I got a 4, and move to the next grade.
By when I graduated from high school, I had convinced myself that math was something I was not “made for”. They were “math and science kids” and I was not one of them. I ended up pursuing a career in, from all things, dancing. I liked dancing, but just as a hobby, more or less in the same manner I liked singing or playing video games. But my father was a self-taught dancer who dreamed of me pursuing the education in dancing he did not (it was mostly about him), and very good at forcing me to do things I did not want to do. I gave up dancing eventually and tried to get into college. At the time, I read a book about the perils of Chilean society and decided I wanted to study sociology (the book author happened to be a sociologist). When I realized sociology had “statistics” in the curriculum, I almost lost hope. I tried hard to learn math for the university admission test on my own (the Chilean “PSU”), but I was failing miserably. Then, I got my first stroke of luck: I just happened to know this guy, Mauro, who gave private math lessons. He was the math tutor of an upper-class friend from high school. It just occurred to me to ask him for math lessons. He got very excited, said yes, and decided to charge me (and my wife) an extremely low fee for the lessons. He literally said: “how much can you pay?” and charged us what we said (he even returned the money back to us at the end). I found out later, that he usually charged ~5 times more what we paid, as he turned out to be one of the best private math instructors in the country, who usually took students from the richest families in Chile at very high fees (he has the largest YouTube channel about math education in Chile now). He also happened to be a politically progressive and extremely generous person. We had around 10-12 study sessions in a ~2.5 months period, just before the university entry exam. I was absolutely flabbergasted by Mauro’s energy, confidence, expertise, and skill. During that time, and for the first time, I realized there was nothing “magical” or “beyond the scope of my skills” about math. Mauro did not want to “show off”, he just wanted to “show us” that we could learn anything. That anyone could learn math given the right instructor, circumstances, and enough time and effort to invest. We did not have anything fancy: old math books, paper, and pencil. That was it. The key was him convincing us that math was accessible, useful, and more importantly, that we could definitively master the contents if we just put the time and effort following his instructions and practicing. This was the first step in a long process of realizing, not just about math, but about any skill or area of expertise, that I could learn pretty much anything. It may take me more time, more effort, a different book, or a different instructor. But it could be accomplished.
During my sociology studies at the Universidad de Chile, I got deeply interested in psychology, neuroscience, and economics. I eventually decided I wanted to do a master’s in economics. I went to talk with a couple of professors in the Economics department at the Universidad de Chile. The conversations I had with them were disheartening. They basically told me that I was way too far behind the required level of mathematical knowledge and skill to get into the program. They also took care to make it sound very difficult and inaccessible for “someone with my background”, as I asked if I could study on my own or take classes over the summer to fill the gaps. They eventually talked me out of applying to the program. They also said that the Public Policy master’s program was “less mathematical” and more suited for someone like me, so I ended up pursuing that program (and eventually graduating at the top of the class).
In 2017, I came to the USA for my PhD in Psychology at UW-Madison with a Fulbright Scholarship for low SES-students. Early on, I got interested in Data Science and Machine Learning. My experience was mixed here. People encouraging me to learn Python and statistics, and people telling me that the class “X” was too hard for me, usually, people related to the Computer Science and Engineering departments. By then, I had enough encouragement and self-confidence to say “Well, let’s try it out and see what happens”. I just started to use all my free time and summers to learn about programming, computer science, mathematics, and anything related to data science and machine learning. It was hard, it took time, but after every iteration learning some new topic, it started to feel easier and easier. I eventually decided to take a graduate-level “mathematics for machine learning” at the engineering department, along with a bunch of graduate students from computer science, mathematics, and engineering. Although I did not take the class for credit, I did attend all classes and did all the homework (I just skipped the midterm and final exam). And that was it: realizing I could do the same exercises, and sometimes even EXPLAIN to graduate students in the engineering and other departments something tricky about a linear algebra problem was eye-opening to me. There is no magic, no special math brain module “enabling” you to learn math.It is about opportunity, good instructors, patience, support, enough resources, time, and effort.
The way I see things in the data science and machine learning space nowadays is like a space dominated by two extremes:
On the one hand, you have people telling others learning Data Science and Machine Learning is dead simple, so simple that you can learn it in 6 months or even 3 months, or by watching their series of twenty short videos, or that math and programming skills are not “really” relevant, so just skip that, it’s a waste of time, no challenges at all! Just conda install this, Shift+Ctrl these cells in my Jupyter Notebook and you are good to apply for a position at Google and Netflix.
On the other hand, you have people telling others that “real” Data Science and Machine Learning require taking this long list of “difficult” math, physics, and computer science classes, or reading cover to cover this stack of books, or publishing at least two papers in NeurIPS, or MA and a PhD in CS/Math/Physics from a “reputable” institution.
I disagree with both perspectives. The first is misleading, creates false hope, trivializes the challenges and effort necessary to learn data science, and encourages people to flood into online courses and bootcamps with the expectation to become a data scientist in 6 months. Many people are wasting time and money following the lead of those individuals. The second is essentially a form of gatekeeping, from people who like to intellectually intimidate others, and who probably need to put others down to protect their socio-economic status and feelings of superiority.
I do not want to say my perspective is a “middle-ground”. It is not. I reject such dichotomy. It is just different. Learning Data Science and Machine Learning is no different than learning about any other subject. Learning about engineering, programming, physical therapy, finance, law, sociology, and any other professional discipline simply requires a mix of opportunity, discipline, resources, good guidance, time, and effort. My philosophy is as follows:put a lot of time and effort learning the fundamentals, and then speed up your learning about more specific subjects. This translates into spending a significant amount of time learning linear algebra, calculus, probability theory, basic algorithms, basic programming, and topics of that kind. If you master the fundamentals, learning more advance or applied topics, like neural networks, graphical models, or randomized experiments, becomes significantly faster and easier. It is essentially like saving money works: the more you invest now, the more you will gain in the future. This approach will probably require a couple of years, but I do not get why such an idea may be controversial. Any person expecting to become an engineer or a psychologist knows that a good 4-6 years must be invested. I am not saying you need 4 years (maybe your background allows you to speed up your process), but I doubt you can get a good data science education in less than 2-3 years. The issue is that many people will make you feel you cannot do it or is too hard, or that you have to be a math person, or a computer science/physics/math major, or just passive-aggressively intimidate you out of the idea. They really like making things sound harder than they are, more alien, more special, and more for a highly selected group of MIT/Stanford/Berkeley graduates. That makes them “feel” smarter, “feel” more special, “feel” more worthy than you are. The other side of the spectrum, as I argued, is no good either.
Source by DesignerRenan
So, what is my revenge all about? I am revenging from all the people who made me doubt myself, of my skill, my intelligence, and my capacity to learn anything, particularly math and science-related subjects. My brain was not missing any “math module” or “special math instinct”. I was just systematically forced out of the topic by people who were too ignorant, too lazy, too unskilled, or too unkind to guide me in the right direction and provide the right support. I like to think that by creating content from the perspective of someone who was told that these topics were beyond its skill and intelligence, can help others in similar positions to regain their confidence and learn what they thought was not possible.