๐Ÿ† ์ž๊ฒฉ์ฆ, ์–ดํ•™

[๋น…๋ฐ์ดํ„ฐ ๋ถ„์„๊ธฐ์‚ฌ] ์‹ค๊ธฐ 7ํšŒ - 3์œ ํ˜• ์ž”์ฐจ์ดํƒˆ๋„

๋ฐ์ดํ„ฐํŒ์Šค 2024. 8. 21. 18:03

 

๋ฌธ์ œ: **train์œผ๋กœ ๋กœ์ง€์Šคํ‹ฑ ํšŒ๊ท€ ์ง„ํ–‰ํ–ˆ์„ ๊ฒฝ์šฐ ์ž”์ฐจ ์ดํƒˆ๋„ (residual deviance)๋ฅผ ๊ณ„์‚ฐํ•˜๋ผ**

# ๋กœ์ง€์Šคํ‹ฑ ํšŒ๊ท€๋ชจํ˜• ์ ํ•ฉ (GLM ์‚ฌ์šฉ) -> ์ดํ•ญ๋ถ„ํฌ์‹œ ๋กœ์ง€์Šคํ‹ฑํšŒ๊ท€
model2 = sm.GLM(y, X, family=sm.families.Binomial()).fit()

# ์ž”์ฐจ ์ดํƒˆ๋„(residual deviance) ๊ณ„์‚ฐ
residual_deviance = model2.deviance
print(residual_deviance)
 

์ž”์ฐจ์ดํƒˆ๋„ ๊ตฌํ•˜๋Š” ๋ฒ•์€ ๋“ค์–ด๋ณธ ์ ๋„ ์—†๋Š”๊ฒŒ ์ด๊ฒŒ ๋ญ๋žŒ? ์ฐพ์•„๋ดค๋‹ค..

'R์˜ ๊ฒฝ์šฐ ํ•ด๋‹น ํ†ต๊ณ„๋Ÿ‰์ด Summary์— footnote๋กœ ์ถœ๋ ฅ๋˜๋Š” ๋ฐ˜๋ฉด ํŒŒ์ด์ฌ์€ ์‚ฌ์ดํ‚ท๋Ÿฐ ๋ฉ”ํŠธ๋ฆญ์Šค๋ฅผ ์ฐพ์•„์„œ ์ถœ๋ ฅํ•ด์•ผ ํ•ด์„œ ์‚ฌ์šฉ ์–ธ์–ด์— ๋”ฐ๋ผ ์œ ๋ถˆ๋ฆฌ๊ฐ€ ๋‹ฌ๋ž๋‹ค.'

[์ถœ์ฒ˜] ๋น…๋ฐ์ดํ„ฐ ๋ถ„์„๊ธฐ์‚ฌ ์‹ค๊ธฐ ํ›„๊ธฐ|์ž‘์„ฑ์ž PDY

 

๊ทธ๋ ‡๋‹ค, ๋ชจ๋ฅด๋Š” ๋‚ด์šฉ์ด๋ผ์„œ ์ฝ”๋“œ๋ฅผ ์™ธ์šฐ๊ธฐ๋กœ ํ–ˆ๋‹ค.

๋กœ์ง€์Šคํ‹ฑ์œผ๋กœ ํšŒ๊ท€ํ•œ ๋‹ค์Œ์—

model.deviance๋ฅผ ํ–ˆ๋”๋‹ˆ ๋กœ์ง€์Šคํ‹ฑ์€ ์ž”์ฐจ ์ดํƒˆ๋„๋ฅผ ๊ตฌํ• ์ˆ˜๊ฐ€ ์—†๋‚˜๋ณด๋‹ค

์ € ํ˜•์‹์„ ์™ธ์šธ ์ˆ˜๋ฐ–์— ์—†๋‹ค..