Related
I want to do logistic regressions for several (n = 30) SNPs (coded as 0,1,2) as predictors and a casecontrol variable (0,1) as an outcome. As few of those rs are correlated, I cannot put all rs# in one model but have to run one at a time regression for each i.e., I cannot simply plus them together in one model like rs1 + rs2 + rs3 and so on....I need to have each regressed separately like below;
test1 = glm(casecontrol ~ rs1, data = mydata, family=binomial)
test2 = glm(casecontrol ~ rs2, data = mydata, family=binomial)
test3 = glm(casecontrol ~ rs3, data = mydata, family=binomial)
While I can run all the above regressions separately, is there a way to loop them together so I could get a summary() of all tests in one go?
I will have to adjust for age and sex too, but that would come after I run an unadjusted loop.
My data head from dput(head(mydata)) for example;
structure(list(ID = 1:6, sex = c(2L, 1L, 2L, 1L, 1L, 1L), age = c(52.4725405022036,
58.4303618001286, 44.5300065923948, 61.4786037395243, 67.851808819687,
39.7451378498226), bmi = c(31.4068751083687, 32.0614937413484,
23.205021363683, 29.1445372393355, 32.6287483051419, 20.5887741968036
), casecontrol = c(0L, 1L, 0L, 1L, 1L, 1L), rs1 = c(1L, 0L, 2L,
2L, 1L, 2L), rs2 = c(2L, 1L, 2L, 0L, 1L, 1L), rs3 = c(1L, 0L,
1L, 2L, 2L, 2L), rs4 = c(1L, 1L, 1L, 1L, 0L, 2L), rs5 = c(1L,
0L, 0L, 0L, 1L, 2L), rs6 = c(1L, 1L, 1L, 1L, 1L, 2L), rs7 = c(0L,
0L, 0L, 0L, 0L, 0L), rs8 = c(0L, 0L, 1L, 0L, 2L, 1L), rs9 = c(0L,
0L, 2L, 1L, 1L, 0L), rs10 = c(2L, 0L, 0L, 2L, 2L, 1L), rs11 = c(0L,
1L, 1L, 0L, 1L, 1L), rs12 = c(1L, 2L, 0L, 1L, 2L, 2L), rs13 = c(0L,
2L, 0L, 0L, 0L, 0L), rs14 = c(1L, 1L, 1L, 1L, 2L, 2L), rs15 = c(1L,
2L, 1L, 1L, 0L, 1L), rs16 = c(0L, 2L, 1L, 2L, 2L, 1L), rs17 = c(0L,
2L, 1L, 1L, 2L, 2L), rs18 = c(1L, 2L, 2L, 1L, 1L, 1L), rs19 = c(1L,
1L, 0L, 1L, 2L, 2L), rs20 = c(2L, 1L, 0L, 2L, 2L, 1L), rs21 = c(1L,
2L, 2L, 1L, 1L, 0L), rs22 = c(1L, 1L, 2L, 2L, 0L, 1L), rs23 = c(2L,
0L, 2L, 1L, 1L, 1L), rs24 = c(0L, 0L, 0L, 2L, 2L, 2L), rs25 = c(2L,
2L, 1L, 1L, 0L, 1L), rs26 = c(1L, 1L, 0L, 2L, 0L, 1L), rs27 = c(1L,
1L, 1L, 1L, 0L, 1L), rs28 = c(0L, 1L, 1L, 2L, 0L, 2L), rs29 = c(2L,
2L, 2L, 2L, 1L, 2L), rs30 = c(0L, 2L, 1L, 2L, 1L, 0L)), row.names = c(NA,
6L), class = "data.frame")```
Probably you want something like this:
lapply(1:30, function(i) glm(as.formula(paste0('casecontrol ~ ', 'rs', i)), data = mydata, family = binomial))
which will execute 30 logistic regressions with the selected predictor.
Instead of hard coding the overall number of predictors, you can use:
sum(grepl('rs', names(mydata))), which will return 30.
You can use tidy function from broom package to get the summary in a tidy format.
purrr::map_dfr(1:30, function(i) data.frame(model = i, tidy(glm(as.formula(paste0('casecontrol ~ ', 'rs', i)), data = mydata, family = binomial))))
or you can do this in a more dynamic way:
names(mydata)[grepl('rs', names(mydata))] -> pred #get all predictors that contain 'rs'
purrr::map_dfr(1:length(pred),
function(i) data.frame(model = i,
tidy(glm(as.formula(paste0('casecontrol ~ ', pred[i])), data = mydata, family = binomial))))
If you want to include another variable, you simply need to adjust the pred vector.
c(pred, paste0(pred, ' + age')) -> pred #interaction between rs drivers and age
or
c(pred, paste0(pred, ' + age + sex')) -> pred #interaction between rs drivers age and sex
you can do something like this
outcome<-mydata %>% select("casecontrol") #select outcome
features <- mydata %>% select("rs1":"rs30") #select features
features_names<-data.frame(colnames(features)) #get feature names
for(i in 1:nrow(features_names)) # loop over length of features_name
{selected=features[,i,drop=FALSE] #get each feature
total <- cbind(selected, response) # combine with outcome
glm.fit <- glm(casecontrol ~ ., data = total, family = binomial("logit"))
summary(glm.fit)
}
I'm new to R. I have a data set
df <- structure(list(schoolid = c(1L, 1L, 1L, 1L, 1L, 1L),
score = c(0L, 10L, 0L, 40L, 42L, 4L),
gender = c(1L, 1L, 1L, 1L, 1L, 1L)),
.Names = c("schoolid", "score", "gender"),
row.names = c(NA, 6L),
class = "data.frame")
for which I have to run a random intercept model to see if there is an impact of the gender on the score across schools. Can anyone kindly explain what is expected from me?
For a sample dataframe:
df <- structure(list(id = 1:19, region.1 = structure(c(1L, 1L, 1L,
1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 4L, 4L, 5L, 5L, 5L
), .Label = c("AT1", "AT2", "AT3", "AT4", "AT5"), class = "factor"),
PoorHealth = c(0L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 1L,
0L, 0L, 0L, 1L, 0L, 1L, 0L, 0L)), .Names = c("id", "region.1",
"PoorHealth"), class = "data.frame", row.names = c(NA, -19L))
I want to subset using the BY command, and hoped somebody may be able to help me.
I want to INCLUDE regions (regions.1) in df that satisfy this condition:
Less than (or equal to) 3 occurrences of '1' in the variable 'PoorHealth'
OR this condition:
Where N (i.e. the respondents in each region) is less than or equal to 6.
If anyone has any ideas to help me, I should be very grateful.
This should work. Dno if there is a cleaner way:
library(data.table)
setDT(df)
qualified_regions = df[,which((sum(PoorHealth==1) <=3 | .N <= 6)),region.1][,region.1]
df[region.1 %in% qualified_regions,]
E: I removed the !-mark because OP changed "EXCLUDE" to "INCLUDE" in the original question.
I have 2 data frames for 2 stacks that gives information about potential emission. One data frame gives the time frame of what hours the system turn on and off for 4 seasons. Each season start on specific date. The 2nd file give me the details of the stack.
I am trying with some sample file to test how to do this and so far I have managed to create a function following stack overflow example that allow me to create a data frame with the dates that I would like and a column with seasons for each date. I am really struggling now with the programming concept to understand how do I combine the 3 data frames to create the output template that I am trying to set up.
To show you an example my sample input are:
Stack_info File:
example seasonal Profile that shows when the system is on or off:
and the output I am after should create data frames for each year in the following format (only the black font and the red text to just explain what the values are):
What is the most difficult I am finding is that my output files for each year will have a unique first Row and the 2nd row will repeat for each pollutant. and from 3rd row the hourly data for all 8760 hours. This need to repeat for the next pollutant.
So far I have managed to create a function that helps me to assign season to each day of the year. For example:
#function to create seasons
d = function(month_day) which(lut$month_day == month_day)
lut = data.frame(all_dates = as.POSIXct("2012-1-1") + ((0:365) * 3600 * 24),
season = NA)
lut = within(lut, { month_day = strftime(all_dates, "%b-%d") })
lut[c(d("Jan-01"):d("Mar-15"), d("Nov-08"):d("Dec-31")), "season"] = "winter"
lut[c(d("Mar-16"):d("Apr-30")), "season"] = "spring"
lut[c(d("May-01"):d("Sep-27")), "season"] = "summer"
lut[c(d("Sep-28"):d("Nov-07")), "season"] = "autumn"
rownames(lut) = lut$month_day
## create date data frame and assign seasons
dates = data.frame(dates =seq(as.Date('2010-01-01'),as.Date('2012-12-31'),by = 1))
dates = within(dates, {
season = lut[strftime(dates, "%b-%d"), "season"]
})
This gives me a dates data frame and my other 2 samples data frames are (as shown in the image):
structure(list(`Source no` = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), Source = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
2L, 2L, 2L, 2L), .Label = c("Stack 1", "Stack 2"), class = "factor"),
Period = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), Day = structure(c(2L,
6L, 7L, 5L, 1L, 3L, 4L, 2L, 6L, 7L, 5L, 1L, 3L, 4L, 2L, 6L,
7L, 5L, 1L, 3L, 4L), .Label = c("Fri", "Mon", "Sat", "Sun",
"Thu", "Tue", "Wed"), class = "factor"), `Spring On` = c(0L,
0L, 0L, 0L, 0L, 0L, 0L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 15L,
15L, 15L, 15L, 15L, 15L, 15L), `Spring Off` = c(23L, 23L,
23L, 23L, 23L, 23L, 23L, 7L, 7L, 7L, 7L, 7L, 7L, 7L, 18L,
18L, 18L, 18L, 18L, 18L, 18L), `Summer On` = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L), .Label = "off", class = "factor"), `Summer Off` = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L), .Label = "off", class = "factor"), `Autumn On` = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L), .Label = "off", class = "factor"), `Autumn Off` = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L), .Label = "off", class = "factor"), `Winter On` = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L,
2L, 2L, 2L, 2L, 2L), .Label = c("0", "off"), class = "factor"),
`Winter Off` = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), .Label = c("23",
"off"), class = "factor")), .Names = c("Source no", "Source",
"Period", "Day", "Spring On", "Spring Off", "Summer On", "Summer Off",
"Autumn On", "Autumn Off", "Winter On", "Winter Off"), class = "data.frame", row.names = c(NA,
-21L)) -> profile
structure(list(SNAME = structure(1:2, .Label = c("Stack 1", "Stack 2"
), class = "factor"), ISVARY = c(1L, 4L), VELVOL = c(1L, 4L),
TEMPDENS = c(0L, 2L), `DUM 1` = c(999L, 999L), `DUM 2` = c(999L,
999L), NPOL = c(2L, 2L), `EXIT VEL` = c(26.2, 22.4), TEMP = c(341L,
328L), `STACK DIAM` = c(1.5, 2.5), W = c(0L, 15L), Nox = c(39,
33.3), Sox = c(15.5, 17.9)), .Names = c("SNAME", "ISVARY",
"VELVOL", "TEMPDENS", "DUM 1", "DUM 2", "NPOL", "EXIT VEL", "TEMP",
"STACK DIAM", "W", "Nox", "Sox"), class = "data.frame", row.names = c(NA,
-2L)) -> stack_info
If anyone could give me any guidance of how I can proceed with the programming part would be really useful as I am just not sure how I can approach this to create separate output files as data frame for year 2010, 2011 and 2012.
The way your data is organised isn't ideal for processing. Maybe you have a look at Hadley Wickhams papar about tidy data.
According to your desired output you need a dataframe with the number of lines equal to the number of hours a specific machine (stack n) is switched on. Therefore I suggest you create a dataframe containing every hour of a given year:
d.out = data.frame(dates = seq(from=as.POSIXct("2010-01-01"), by=3600, to= as.POSIXct("2010-12-31")))
d.out$year = as.numeric(format(d.out$dates, "%Y"))
d.out$month = as.numeric(format(d.out$dates, "%m"))
d.out$day = as.numeric(format(d.out$dates, "%d"))
d.out$hour = as.numeric(format(d.out$dates, "%H"))
d.out$weekday = as.character(format(d.out$dates, "%a"))
d.out$doj = as.numeric(format(d.out$dates, "%j"))
d.out$season = "Winter"
d.out$season[d.out$doj >= 75 & d.out$doj < 121] = "Spring"
d.out$season[d.out$doj >= 121 & d.out$doj < 271] = "Summer"
d.out$season[d.out$doj >= 271 & d.out$doj < 312] = "Autumn"
The goal is to join this dataframe with your profile dataframe. Before joining, the profile-df has to be rearranged:
library(dplyr)
library(tidyr)
profile_new =
profile %>%
gather(season, hour, -c(`Source no`, Source, Period, Day)) %>%
extract(season, c("season", "status"), "(\\w+?)\\s(\\w+)") %>%
filter(hour != "off") %>%
mutate(Day = as.character(Day), hour=as.numeric(hour)) %>%
spread(status, hour)
Now it's easy to join the three dataframes to put together all the information you need to create your output:
d.out %>%
inner_join(profile_new, by=c("weekday"="Day", "season"="season")) %>%
group_by(Source, dates, year, day, weekday, season, hour) %>%
summarise(status = any(hour >= On & hour <= Off)) %>%
inner_join(stack_info, by=c("Source"="SNAME")) %>%
mutate(Nox = ifelse(status, Nox, 0),
Sox = ifelse(status, Sox, 0)) %>%
arrange(Source, year, dates, hour) %>%
select(Source, year, day, weekday, season, hour, `EXIT VEL`, TEMP, `STACK DIAM`, W, Nox, Sox)
Obviously it's not quite the format you posted. From here you could write your dataframe to a csv (stack by stack by using append = TRUE).
So, I have a data frame with several continuous variables and several dummy variables. The survey that this data frame comes from uses 6,7,8 and 9 to denote different types of non-response. So, I would like to replace 6,7,8 and 9 with NA whenever they show up in a dummy variable column but leave them be in the continuous variable column.
Is there a concise way to go about doing this?
Here's my data:
> dput(head(sfsuse[c(4:16)]))
structure(list(famsize = c(3L, 1L, 2L, 5L, 3L, 5L), famtype = c(2L,
1L, 2L, 3L, 2L, 3L), cc = c(1L, 1L, 1L, 1L, 1L, 1L), nocc = c(1L,
1L, 1L, 3L, 1L, 1L), pdloan = c(2L, 2L, 2L, 2L, 2L, 2L), help = c(2L,
2L, 2L, 2L, 2L, 2L), budget = c(1L, 1L, 1L, 1L, 2L, 2L), income = c(340000L,
20500L, 0L, 165000L, 95000L, -320000L), govtrans = c(7500L, 15500L,
22000L, 350L, 0L, 9250L), childexp = c(0L, 0L, 0L, 0L, 0L, 0L
), homeown = c(1L, 1L, 1L, 1L, 1L, 2L), bank = c(2000L, 80000L,
25000L, 20000L, 57500L, 120000L), vehval = c(33000L, 7500L, 5250L,
48000L, 8500L, 50000L)), .Names = c("famsize", "famtype", "cc",
"nocc", "pdloan", "help", "budget", "income", "govtrans", "childexp",
"homeown", "bank", "vehval"), row.names = c(NA, 6L), class = "data.frame")
I'm trying to subs in NA for 6,7,8 and 9 in columns 3:7 and column 11. I know how to do this one column at a time by the column names:
df$name[df$name %in% 6:9]<-NA
but I would have to do this for each column by name, is there a concise way to do it by column index?
Thanks
This function should work
f <- function(data,k) {
data[data[,k] %in% 6:9,k] <- NA
data
}
Now at the console:
> for (k in c(3:7,11)) { data <- f(data,k) }