How to use apply function instead of for loop if y

1st DF:

t.d
  V1 V2 V3 V4
1  1  6 11 16
2  2  7 12 17
3  3  8 13 18
4  4  9 14 19
5  5 10 15 20


names(t.d) <- c("ID","A","B","C")

t.d$FinalTime <- c("7/30/2009 08:18:35","9/30/2009 19:18:35","11/30/2009 21:18:35","13/30/2009 20:18:35","15/30/2009 04:18:35")

t.d$InitTime <- c("6/30/2009 9:18:35","6/30/2009 9:18:35","6/30/2009 9:18:35","6/30/2009 9:18:35","6/30/2009 9:18:35")

>t.d
  ID  A  B  C           FinalTime          InitTime
1  1  6 11 16  7/30/2009 08:18:35 6/30/2009 9:18:35
2  2  7 12 17  9/30/2009 19:18:35 6/30/2009 9:18:35
3  3  8 13 18 11/30/2009 21:18:35 6/30/2009 9:18:35
4  4  9 14 19 13/30/2009 20:18:35 6/30/2009 9:18:35
5  5 10 15 20 15/30/2009 04:18:35 6/30/2009 9:18:35

2nd DF:

> s.d
   F  D  E                Time
1  10 19 28  6/30/2009 08:18:35
2  11 20 29  8/30/2009 19:18:35
3  12 21 30  9/30/2009 21:18:35
4  13 22 31 01/30/2009 20:18:35
5  14 23 32 10/30/2009 04:18:35
6  15 24 33 11/30/2009 04:18:35
7  16 25 34 12/30/2009 04:18:35
8  17 26 35 13/30/2009 04:18:35
9  18 27 36 15/30/2009 04:18:35

Output to be:

From DF "t.d" I have to calculate the time interval for each row between "FinalTime" and "InitTime" (InitTime will always be less than FinalTime).

Another DF "temp" from "s.d" has to be formed having data only within the above time interval, and then the most recent values of "F","D","E" have to be taken and attached to the 'ith' row of "t.d" from which the time interval was calculated.

Also we have to see if the newly formed DF "temp" has the following conditions true:

here 'j' represents value for each row:

if(temp$F[j] < 35.5) + (temp$D[j] >= 100) >= 1)
{
  temp$Flag <- 1
} else{
  temp$Flag <- 0
}

Originally I have 3 million rows in the dataframe and 20 columns in each DF.

I have solved the above problem using "for loop" but it obviously takes 2 to 3 days as there are a lot of rows.

(Also if I have to add new columns to the resultant DF if multiple conditions get satisfied on each row?)

Can anybody suggest a different technique? Like using apply functions?

My suggestion is:

use lapply over row indices
handle in the function call your if branches
return either your dataframe or NULL
combine everything with rbind

by replacing lapply with mclapply from the 'parallel' package, your code gets executed in parallel.

resultList <- lapply(1:nrow(t.d), function(i){
do stuff
if(condition){
    return(df)
}else{
    return(NULL)
}
resultDF <- do.call(rbind, resultList)