“If you don’t know what to do, go find something and kill it.”
Or so once said Field Marshall Rommel. Now I’m pretty sure that herr Rommel, by virtue of having died more than 40 years before personal computers, did not know much about programming or data science. Even so, his wisdom will come in handy today.
So…Welcome to my programming and data science blog. I’ll start with mission statement, a little bit about my background and finally, my gameplan.
My first intent is to create a set of notes as tutorials as I learn more about machine learning. I will try to make it as easy to follow as possible, as I plan to refer to them as well in the future.
It is also to document the struggles, the dead ends, the frustrations, the joys and the tears.I expect detours, false starts, dead ends, etc so that I may look back 10 years from now and laugh at my early tribulations. Hopefully, somebody else also takes something out of it, but I have to admit this is a selfish pursuit.
I have a degree in math, and while I’ll try my best to explain things, it’s important that I keep momentum so I might handwave a few details. Mea culpa.
Now the Gameplan. The very first project will be this kaggle competition:
My main reason was that the dataset is small and I only have very modest hardware (about comparable to what except with less RAM and VRAM).
Roughly, my attempt will follow this structure:
0. Descriptive phase, where I explore the dataset
1. Literature review
2. Feature engineering, where I seek to rearrange features and create new ones .
3. Model(s) selection/creation
4. Training and testing
Thus far, nothing too revolutionary. If you are minimally interested in machine learning, you have seen this plan somewhere. Unlike a roadtrip, the steps aren’t necessarily discrete and sequential , I fully expect them to overlap and for me to go back from time to time.
So that’s it for today. Tomorrow: phase zero.