Is chess like learning data science?
Or How is learning like learning?
Chess is one of my passions. I play in tournaments, play online, teach the game, and organize and run tournaments. Learning Data Science is also a passion of mine and sometimes I twitch when the same neural pathways are accessed, so to speak.
When I talk about learning chess I’m not describing how the pieces move or the rules of the venerable game. I am discussing an intimate understanding of the game, the nuances of different positions and why a move was made and what does the move of my opponent threaten? What to do next and why to do it?
When you start out, you learn the language of chess. This is called the algebraic notation and allows one to record a game and to walk through it afterwards. Its fairly easy to learn and base on the grid of the board of 8 ranks and 8 files, each square has a name i.e e4, h6, b2.
A game is broke down into simple three phases. At least the names are simple enough: Opening, Middlegame, Endgame.
In the Opening, you develop your pieces off the back rank, control the center with pawns or pieces, move your king to safety by castling. Work to get a good position.
The Middlegame usually begins when the rooks have been connected meaning all pieces have been developed and the king has castled. Maneuvering pieces to control the center and make a plan to attack the king are part of the middlegame modus operandi. Tactics abound in the Middlegame and recognizing patterns starts to dominate. Above all, the purpose of the Middlegame is to transition into a winning endgame.
In the Endgame, the King comes out of hiding and is a fighting piece. Pawn play is critical as one wrong move could erase all the previous ones and cost the game. Rooks support pawns or invade to increase chances to create a passed pawn in order to promote to a Queen.
Now take a moment, Think about one of your Data Science projects. How is your own journey of learning Data Science similar?
In learning Data Science, you should have a knowledge of a programming language normally either Python or R.
You learn a process, a way to proceed in the analysis of said data. Three common processes are CRISP-DM, KDD, or OSEMN.
You learn statistics, data visualization, feature creation, transformations, data cleaning, modeling, algebra, calculus, and much more. You don’t always use these tools but you know how to use them and when to use them. They are in the tool box ready to be called up on need. Similar to chess tactics. Some are called removal of the defender, forks, skewers, deflection and probably the most famous, the sacrifice.
You learn the different algorithms to accomplish the analysis of your data. You use different algorithms depending on what the data dictates as you use the different tactics just as the position on the board dictates.
Learning a process is akin to an opening or possible the rules of the game. (I know the process is a framework and rules are rules but its a rough analogy) As to how to start the game which defines your strategy and you have a general idea for next steps.
Feature Creation is the creative way similar to creating a sound attack and knowing that it will be beneficial later. Chess has been called an art at times as well as science and pure logic. This statement clearly defines Data Science in my experience as it is art and science as well.
One can learn the techniques, and algorithms similar to tactics or chess openings but not understanding the measures or signals of your algorithm is like learning the first 10 moves of an opening and finding yourself in a complicated and unfamiliar middlegame.
Pattern recognition is learned by practice and doing and studying. It is a crucial building block to being a competent player in chess and Data Science. The more you do and understand the pattern recognition becomes built in, it becomes intuition.
Current World Champion Magnus Carlsen of Norway, once said he knows within a few seconds what the best move is but spends his time checking (2 to 5 or more moves) ahead to see if he missed anything. This intuition was learned by study of current opening trends but also by studying many great games of the past. Here is a fascinating video of his recollection being tested.
The comparison to chess is definitely overused as you hear in many sporting events “we sure do have a chess game now”. This example is not a perfect analogy but there are many similarities which I see and I hope you do as well.