# STAT3530 -- variable selection

site <- 'https://tdruiz-stat3530.share.connect.posit.cloud/_data/'
load(url(paste0(site, 'pollution.RData')))

plot(mortality ~ log(so2), data = pollution, pch = 16, col = 'grey35')

# ---------------------------------------------------------------------------
# 1. simple linear regression
# ---------------------------------------------------------------------------

fit0 <- lm(mortality ~ log(so2), data = pollution)

summary(fit0)$coefficients['log(so2)', ]
confint(fit0)['log(so2)', ]

# ---------------------------------------------------------------------------
# 2. your model -- adjust for exactly six
# ---------------------------------------------------------------------------
#   precip  humidity  jantemp  julytemp
#   over65  hhsize  educ  housing  density  nonwhite  whitecol  poor
#   log(hc)  log(nox)

fit <- lm(mortality ~ log(so2) + ..., data = pollution)

summary(fit)$coefficients['log(so2)', ]
confint(fit)['log(so2)', ]

choose(14, 6)

# ---------------------------------------------------------------------------
# 3. upload
# ---------------------------------------------------------------------------

me <- '...'

saveRDS(fit, file = paste0('model-', me, '.rds'))