Data Science, the old school way
To me, software engineering is a passion and a feeling of selffulfillment which, when coupled with fruitful projects giving way to advancements in society and humanity as a whole, flares a deeply rooted passion to learn learn learn! This is why, when I first read up on Data Science, I knew that I had honed in on my next target to devour and assimilate into my neural network (yes, the biological one!)
But as I started reading through the hundreds of howto’s, blogs, MOOCS, video tutorials, online books, PDFs and the rest, I only found more frustration with missing pieces to a larger and more obscure logical puzzle that was defiant and unyielding. A feeling of unrest and concern surfaced my mind as I delved more deeply into the subject getting halfbaked and limited understanding allowing me to build only by following “handholding” step by step instructions. This was not engineering! Trial after trial the matter finally became clear to me — The topic of Data Science strictly refuses to truly enter a mind until it is imbued with proper mathematical/statistical foundations!
As my reading on Data Science progressed, I would read or hear terms that I knew were key to understanding how Neural Networks worked, but it was just half baked! I started seeing this trend of people seizing the opportunity to piggy back off the Data Science craze and authoring reads and making full fledged online “Data Camps” that in the end will always avoid the real mechanics behind the subject. They breeze through the parts where mathematical foundations are necessary and move on as if it was a side matter and not required to build reliable models. Take this snippet from a free online book for Neural Networks as an example,
“On a related note, the mathematical requirements to read the book are modest. There is some mathematics in most chapters, but it’s usually just elementary algebra and plots of functions, which I expect most readers will be okay with. I occasionally use more advanced mathematics, but have structured the material so you can follow even if some mathematical details elude you.”
Reading through only the second half of Chapter 1: “ We’ll also define the gradient of C to be the vector of partial derivatives”, “We denote the gradient vector by ∇C” and“Then Equation (9) tells us that ΔC≈−η∇C⋅∇C=−η∥∇C∥2.”, all prove otherwise and goes against being “elementary.”
This is why I found myself constantly referring back to tutorials that explained the topic of gradient descent, limits, derivatives, vectors, statistics then go back to continue my reading. But this style of learning left me with a feeling of despair resonating from my logical side whispering, “It’s not enough, I still want to know WHY and HOW!”
Then, one day as I was cleaning up my closet, my eyes suddenly fell on my old college Single Variable Calculus book by James Stewart. Followed by a sudden dramatic stare and a sense of joy from a reunion with an old friend, an innate feeling told me that an epiphany is about to occur! I opened the book  chapter 1  and lo and behold! “This…is it. It’s…it’s all there! IT’S ALL THERE!” Really, all the missing pieces and loose ends were so finely andreliably described that it left no room for confusion. Yes, I do agree that only after having gone through all the Data Science tutorials that Calculus made so much more sense  but it was still a critical piece to the big Data Science puzzle.
The point I’m driving here is that I feel that there are many fields in I.T. that are slowly being driven by marketing objectives and quick responses to technology hypes rather than the classic RT*M way of deeply rooted “old school” developers. Head hunters are scouring to find anyone that can “fit the description” screening people through checkoff lists and “how many years of” rather than searching for minds withchemical X which in turn forces future Data Scientist to prepare to become real engineers. The pressure coming down from this flawed style of talent selection is slowly giving rise to a batch of cooky cutter engineers that sparks worry given the way the entire field of I.T. ties in together. Security, reliability, scalability and general advancements in I.T. are all negatively effected by allowing the field to grow without well founded roots. Take how Android programming tutorials started as a classic example. The push and drive behind most of the videos, MOOCs, and tutorials was to quickly jump on the demand and make things seem as easy as possible to drive more readers and not towards making strong android engineers. This finally resulted in poor android engineers that designed poorly designed apps making way for many security and stability issues. I believe this is the same trend that is taking place in the Data Science field.
In summary, if you really want to understand Data Science in a way that will allow you to have a positive impact and pave the way for your future in this ever growing field then I strongly suggest studying the basics really well  along with all the MOOCs, videos and tutorials that are still key in understand Data Science and Machine Learning.
Learning Data Science the right way
2 posts
• Page 1 of 1
Why this so hard to find a job in this area? Because a lot of employers have unrealistic expectations in data science. Pretty much a job description is expectations for an entire dev team. Most employers expect to have a data scientist to have a Ph.D. and be a great programmer which are two different people. If you have no clue what the is data since overall. Suggest you visit this sdsclub website


2 posts
• Page 1 of 1
Who is online
Users browsing this forum: No registered users and 4 guests