본문 바로가기

카테고리 없음

confidence interval vs prediction interval

728x90
반응형

https://statisticsbyjim.com/hypothesis-testing/confidence-prediction-tolerance-intervals/

 

Confidence Intervals vs Prediction Intervals vs Tolerance Intervals - Statistics By Jim

Confidence intervals are common in statistics but other intervals can be more useful. Learn when to use confidence, prediction, and tolerance intervals.

statisticsbyjim.com

confindence interval : 지정된 신뢰구간 내에서 예측치들의 평균값이 위치할 범위

prediction interval : 지정된 신뢰구간 내에서 예측치들이 존재할 범위

 

x=sort(rnorm(100))
y=sort(rnorm(100))

df=data.frame(x=x,y=y,stringsAsFactors=F)
model1=lm(y~x,data = df)
pred_interval=predict(model1,newdata=df,interval='prediction',level=0.95)

# Prediction interval
plot(x,y)
lines(x,pred_interval[,1],lwd=2)
lines(x,pred_interval[,2],lwd=2,col='red')
lines(x,pred_interval[,3],lwd=2,col='skyblue')

# Confindence interval
conf_interval=predict(model1,newdata=df,interval='confidence',level=0.95)
lines(x,conf_interval[,2],lwd=2,col='purple')
lines(x,conf_interval[,3],lwd=2,col='green')

# Legend
legend('topleft',lwd=5,col=c('red','skyblue','purple','green'),
legend = c('Pred_low','Pred_top','Conf_low','Conf_top'))

728x90
반응형