library(data.table)

file.in <- "path/to/your/file.csv"
DT.accidents <- fread(file.in)

## Have a look at the different DRUNK_DR values
DT.accidents[, table(DRUNK_DR)]
## Nine?? Really?  

DT.accidents[DRUNK_DR == 9]

## Anyway, to sum up by state and drunk drivers, assuming one row of data is one accident, you can simply use: 

DT.accidents[, .N, by=list(STATE, DRUNK_DR)]

## If you want to ignore cases with zero drunk drivers, filter those out
DT.drunks <- DT.accidents[DRUNK_DR > 0, .N, by=list(STATE, DRUNK_DR)]

## You can reshape it too, if you'd like

library(reshape2)
DT.drunks <- as.data.table(dcast(DT.drunks, STATE ~ DRUNK_DR, value="N"))

添加状态名称

State Names, accorindg to 
ftp://ftp.nhtsa.dot.gov/FARS/FARS-DOC/USERGUIDE-2007.pdf

## start with the built in variable 'state.name' (no "s")
state_names <- state.name[1:50]
## Add in territories
state_names <- sort(c(state_names, "District of Columbia", "Puerto Rico", "Virgin Islands"))
## Create index numbers that match what is shown in the file
state_number <- setdiff(1:56, c(3, 7, 14))
## Create a data.table for joining
DT.states <- data.table(state_number=state_number, state_names=state_names)

## Join in the info
setkey(DT.states, "state_number")
setkey(DT.accidents, "STATE")
DT.accidents[DT.states, STATE_NAMES := state_names]

## Now you can reshape, but include the names
DT.drunks <- DT.accidents[DRUNK_DR > 0, .N, by=list(STATE, STATE_NAMES, DRUNK_DR)]

## You can reshape it too, if you'd like
DT.drunks <- as.data.table(dcast(DT.drunks, STATE + STATE_NAMES ~ DRUNK_DR, value="N"))

现在......至于那起九人酒后驾车的事故。

DT.accidents[DRUNK_DR == 9]

谷歌搜索：“蒙大拿州2007年5月19日交通事故”第一个结果是http://www.city-data.com/accidents/acc-Browning-Montana.html
它有这样一条信息
布朗宁2007年致命车祸和道路交通事故（沉船）名单：2007年5月19日下午05：55，美国2号公路，Sr-464，纬度：48.555692，隆：-113.010247，车辆：1、死亡事故：1、酒后驾车：数据不一致