728x90
๋ฐ˜์‘ํ˜•
๐Ÿ’ก
๋ชจ๋ธ๋ง(Modeling) : ์ˆ˜ํ•™/ํ†ต๊ณ„์ ์ธ ๋ฐฉ๋ฒ•์„ ์ด์šฉํ•˜์—ฌ ๋ฐ์ดํ„ฐ๋ฅผ ํ•ด์„ํ•˜๋Š” ๊ณผ์ •.
  • ๊ทผ๋ณธ์ ์ธ ๊ฐœ๋…์€ ์ˆ˜ํ•™์ ์ธ ๋ชจ๋ธ๊ณผ ํ†ต๊ณ„์ ์ธ ๋ชจ๋ธ์˜ ๊ฐœ๋…์—์„œ ๋‚˜์˜จ๋‹ค.
Mathematical model - Wikipedia
A mathematical model is a description of a system using mathematical concepts and language. The process of developing a mathematical model is termed mathematical modeling. Mathematical models are used in the natural sciences (such as physics, biology, earth science, chemistry) and engineering disciplines (such as computer science, electrical engineering), as well as in non-physical systems such as the social sciences (such as economics, psychology, sociology, political science).
https://en.wikipedia.org/wiki/Mathematical_model
Statistical model - Wikipedia
A statistical model is a mathematical model that embodies a set of statistical assumptions concerning the generation of sample data (and similar data from a larger population). A statistical model represents, often in considerably idealized form, the data-generating process. A statistical model is usually specified as a mathematical relationship between one or more random variables and other non-random variables.
https://en.wikipedia.org/wiki/Statistical_model
  • ๋ฐ์ดํ„ฐ ์‚ฌ์ด์–ธ์Šค์—์„œ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ๋ธ์€ ํ†ต๊ณ„ ๋ชจ๋ธ ๊ทธ ์ž์ฒด์ผ ์ˆ˜๋„ ์žˆ๊ณ , ๋ฐ์ดํ„ฐ๋ฅผ ๋ถ„์„ํ•˜๋Š” ์–ด๋–ค ์‹œ์Šคํ…œ ๊ทธ ์ž์ฒด์ผ ์ˆ˜๋„์žˆ๋‹ค.
  • ๋ชจ๋ธ์€ ํ•˜๋ ค๊ณ  ํ•˜๋Š” task์— ๋”ฐ๋ผ ์—ฌ๋Ÿฌ๊ฐ€์ง€ ์ด๋ฆ„์œผ๋กœ ๋ถˆ๋ฆฐ๋‹ค. e.g. ์˜ˆ์ธก ๋ชจ๋ธ, ์ƒ์„ฑ ๋ชจ๋ธ, ์š”์•ฝ ๋ชจ๋ธ, โ€ฆ
  • ๋ฐ์ดํ„ฐ ์‚ฌ์ด์–ธ์Šค๋Š” ์ฃผ๋กœ ์˜ˆ์ธก ๋ชจ๋ธ์ด ์‚ฌ์šฉ๋œ๋‹ค.
  • ์˜ˆ์ธก ๋ชจ๋ธ์ด๋ž€, ์ฃผ์–ด์ง„ ๋ฐ์ดํ„ฐ์˜ ํŒจํ„ด์„ ํŒŒ์•…ํ•ด์„œ ๋ฏธ๋ž˜์˜ ๋ฐ์ดํ„ฐ์˜ ํŒจํ„ด์„ ์˜ˆ์ธกํ•  ์ˆ˜ ์žˆ๋Š” ์‹œ์Šคํ…œ์„ ๋งํ•œ๋‹ค.
  • ๋ชจ๋ธ๋ง์—๋Š” ํ†ต๊ณ„์  ๋ชจ๋ธ๋ง๊ณผ ๋จธ์‹ ๋Ÿฌ๋‹(Machine Learning) ๋ฐฉ๋ฒ•์ด ์ฃผ๋กœ ์‚ฌ์šฉ๋œ๋‹ค.

Statistical Modeling

๐Ÿ’ก
ํ†ต๊ณ„์  ๋ชจ๋ธ๋ง : ๋ฐ์ดํ„ฐ๊ฐ€ ์ •์˜๋˜๋Š” ๊ณต๊ฐ„(S, sample space)์™€ ๊ทธ ๊ณต๊ฐ„์—์„œ์˜ ํ™•๋ฅ  ๋ถ„ํฌ(P, probability distribution on S)๊ฐ€ ์žˆ๋‹ค๊ณ  ํ•  ๋•Œ, ์–ด๋–ค ๋ฐ์ดํ„ฐ๋Š” S์—์„œ์˜ P๋ฅผ ํ†ตํ•ด์„œ ๋งŒ๋“ค์–ด์กŒ๋‹ค๊ณ  ๊ฐ€์ •ํ•˜๊ณ  ๋ฐ์ดํ„ฐ๋ฅผ ์ž˜ ๊ธฐ์ˆ ํ•˜๋Š” P๋ฅผ ์ฐพ๋Š” ๊ณผ์ •.
  • ๋ณดํ†ต P๋Š” parameter๋“ค์— ๋Œ€ํ•ด์„œ ์ •์˜๋œ๋‹ค. /์ˆ˜ํ•™
  • parameter๋ž€ ์–ด๋–ค ํ†ต๊ณ„ ์ง‘๋‹จ์„ ๊ธฐ์ˆ ํ•˜๋Š” ์ธก์ •๋œ ๊ฐ’์ด๋‹ค. e.g. ํ‰๊ท (mean, ฮผ\mu๏ปฟ), ํ‘œ์ค€ํŽธ์ฐจ(standard deviation, ฯƒ\sigma๏ปฟ) ...
  • ์ ์ ˆํ•œ parameter๋ฅผ ์ฐพ๊ฒŒ ๋˜๋ฉด ์ž˜ ๊ธฐ์ˆ ํ•˜๋Š” P๋ฅผ ์ฐพ์„ ์ˆ˜ ์žˆ๋‹ค.
  • ์ด๋Ÿฌํ•œ parameter๋ฅผ ์ฐพ์•„๊ฐ€๋Š” ๊ณผ์ •์ด๋ผ๊ณ  ๋ณผ ์ˆ˜ ์žˆ๋‹ค.

Machine Learning Modeling

๐Ÿ’ก
๋จธ์‹ ๋Ÿฌ๋‹ ๋ชจ๋ธ๋ง : ์ฃผ์–ด์ง„ ํ•™์Šต ๋ฐ์ดํ„ฐ๋กœ ์–ป์€ ์ •๋ณด๋กœ ํ•™์Šตํ•˜์ง€ ์•Š์€ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•ด์„œ ์˜ˆ์ธก(๋˜๋Š” ์ถ”๋ก )์„ ํ•˜๋Š” ๊ณผ์ •
  • ๋จธ์‹ ๋Ÿฌ๋‹ ๋ชจ๋ธ์€ ์—ฌ๋Ÿฌ๊ฐ€์ง€ ์ข…๋ฅ˜๊ฐ€ ์žˆ๋‹ค.
  • ์–ด๋–ค task๋ฅผ ์ˆ˜ํ–‰ํ•˜๋А๋ƒ์— ๋”ฐ๋ผ์„œ ์ ํ•ฉํ•œ ๋ชจ๋ธ๋“ค์ด ์žˆ๋‹ค. e.g. ๋ถ„๋ฅ˜ - Logistic Regression, Naive Bayes, ... ํšŒ๊ท€ - Linear Regression, Random Forest, ...
  • ํ†ต๊ณ„์  ๋ชจ๋ธ๋ง ์ฒ˜๋Ÿผ ์ ์ ˆํ•œ Parameter(๋˜๋Š” weight)๋ฅผ ์ฐพ๋Š” ๊ณผ์ •์ด๋ผ๊ณ  ๋ณผ ์ˆ˜ ์žˆ๋‹ค.
  • ํ•™์Šต ๋ฐฉ๋ฒ•์„ ๊ฒฐ์ •ํ•˜๋Š” parameter์ธ hyper-parameter์˜ ์˜ํ–ฅ์„ ๋ฐ›๋Š”๋‹ค.

์š”์•ฝ

  • ํ†ต๊ณ„์  ๋ชจ๋ธ๋ง์€ ๋ฐ์ดํ„ฐ๋ฅผ ํ•ด์„ํ•  ์ˆ˜ ์žˆ๋Š” ํ™•๋ฅ  ๋ถ„ํฌ๋ฅผ ์ฐพ์•„๊ฐ€๋Š” ๊ณผ์ •์ด๋‹ค. ๊ทธ ํ™•๋ฅ ๋ถ„ํฌ๋Š” parameter๋ฅผ ํ†ตํ•ด ๊ฒฐ์ •๋œ๋‹ค.
  • ๋จธ์‹ ๋Ÿฌ๋‹ ๋ชจ๋ธ๋ง์€ ์ฃผ์–ด์ง„ ํ•™์Šต๋ฐ์ดํ„ฐ๋กœ ์ •๋ณด๋ฅผ ํ•™์Šตํ•ด์„œ ํ•™์Šตํ•˜์ง€ ์•Š์€ ๋ฐ์ดํ„ฐ๋ฅผ ํŒ๋‹จํ•˜๋Š” ๊ธฐ์ค€์„ ์„ค์ •ํ•˜๋Š” ๊ณผ์ •, ๊ทธ ํŒ๋‹จํ•˜๋Š” ๊ธฐ์ค€์€ parameter๋ฅผ ํ†ตํ•ด ๊ฒฐ์ •๋œ๋‹ค.
  • ๋ชจ๋ธ๋ง์ด๋ผ๋Š” ๊ฒƒ์€ ์–ด๋– ํ•œ parameter๋ฅผ ์ฐพ์•„๊ฐ€๋Š” ๊ณผ์ •์ด๋‹ค.


Hands-on

  1. Mathematical Model๊ณผ Statistical Model์— ๋Œ€ํ•œ ์˜๋ฌธ ์œ„ํ‚คํ”ผ๋””์•„๋ฅผ ์ฝ๊ณ , ๋‚˜๋ฆ„๋Œ€๋กœ ์ƒ๊ฐํ•œ ์ •์˜๋ฅผ ์ •๋ฆฌํ•ด๋ณด์„ธ์š”.
  1. ๋Œ€ํ‘œ์ ์ธ ๋จธ์‹ ๋Ÿฌ๋‹ ๋ชจ๋ธ๋“ค์„ 3๊ฐ€์ง€ ์ •๋„ ์ฐพ์•„์„œ ์ •๋ฆฌํ•ด๋ณด์„ธ์š”.

Uploaded by N2T

728x90
๋ฐ˜์‘ํ˜•

+ Recent posts