R回归循环

Question

I am attempting to write an R script that will run through a regression for every 1000 rows of data.

I would also like to store the output into a csv file.

Data Link: https://drive.google.com/open?id=0B-W2Dyw4QjoMQ0F1ZFhsdWJORHM

Any help would be great!

 z <- read.csv("Merge.csv", sep=",", header=TRUE, stringsAsFactors=FALSE)


co<-matrix(NA, nrow=44, ncol = 7)

## Regression for mkt returns and 7 variables 
for (i in 5:44){
  cat(i)
  temp <- lm(ret~earningsyld+book2mkt+leverage+corptax+financials+Momentum+lnmktcap, na.action="na.omit", data = z[1000*(i-1)+1:1000*i,])
  co[i,]<- summary(temp)$coefficient [2:8,1]
}

Answer 1

# Built-in example data
data("mtcars")
mtcars <- rbind(mtcars,mtcars,mtcars,mtcars,mtcars,mtcars,
                mtcars,mtcars,mtcars,mtcars,mtcars)

# Let's pretend that instead of 1000 rows it's 10, so that I can use builtin data to show you

co<-matrix(NA, nrow=35, ncol = 1)

## Regressions 
for (i in 1:floor(nrow(mtcars)/10)){
  cat(i)
  temp   <- lm(mpg~cyl, na.action="na.omit", data = mtcars[((i-1)*10)+1:(i*10),])
  co[i,] <- summary(temp)$coefficients[2:2,1]
}

head(co)

  [,1] [1,] -1.696939 [2,] -3.461047 [3,] -2.572180 [4,] -2.651129 [5,] -3.023638 [6,] -2.712391 

I believe there was a problem with your indexing as well as an incorrect spelling of the coefficients element of the summary object. Any other issues would be data issues, so if you still have problems after this be sure to provide the data.

For example if a problem in the data (singularity, all missing values, no identifying variation, etc) causes the inestimitability of a coefficient then it will change the size of the returned summary and break your loop. The only solution to that would be to clean your data or change your model to avoid using bad data.