Loop for regression over multiple factors - regression

Question

I am struggling to get a loop to run several regressions and store the coefficients and intercepts. I have a data similar as this:
data <- data.frame(y = rnorm(10), x1 = rnorm(10)*2, ID = c(rep(1,10), rep(2,10)), group = c(rep(3,5), rep(4,5)))
Where ID and group are factors, therefore:
data$ID <- as.factor(data$ID)
data$group <- as.factor(data$group)
So far I tried 2 approaches.
First I did the following:
for (i in unique(data$ID)){
for (j in unique(data$group)){
fit <- glm(y ~ x1, data=data[data$ID == i & data$group == j, ])
}
}
Afterwards I did the following:
myfun <- function(data) {
step(glm(y ~ x1, data = data), trace=0)
}
fcomb <- unique(data[,c("ID","group")])
mod <- list()
for(i in 1:nrow(fcomb)) {
mod <- c(mod,list(myfun(subset(data,ID==fcomb$ID[i] & group==fcomb$group[i]))))
}
In the end I would like to have a dataset in which for each ID and group I would have the intercept and the beta for the effect of x1 in y.
When I performed the second strategy I got something, but the betas and the intercepts are the same (which is totally impossible) and I still don't know how to store the values.

Answer 1

set.seed(1839)
data <- data.frame(
y = rnorm(10),
x1 = rnorm(10) * 2,
ID = c(rep(1, 10), rep(2, 10)),
group = c(rep(3, 5), rep(4, 5))
)
grid <- expand.grid(ID = unique(data$ID), group = unique(data$group))
results <- lapply(1:nrow(grid), function(x) {
lm(y ~ x1, data[data$ID == grid[x, 1] & data$group == grid[x, 2], ])$coef
})
results <- t(do.call(cbind, results))
results <- cbind(grid, results)
results
Returns:
ID group (Intercept) x1
1 1 3 -0.454072247 1.0295731
2 2 3 -0.454072247 1.0295731
3 1 4 0.007800405 -0.1832663
4 2 4 0.007800405 -0.1832663