How to extract block of rows in R - r

Question

here is a example of my data frame (the original has ~ 10 000 rows). I would like to extract blocks of row based on VariableC. I only want to keep rows between FALSEs. But only "blocks" with a minimum number of rows of 10 (randomly located in the data frame) and discard the others. In other words, I want to split my data frame into sub data frames (i.e. block of rows). An alternative would be to create a new column with each block having an individual number or letter. The end goal is to plot (regression) VariableA and VariableB for each block and extract the regression and slope coefficients of each block. I know how to do the last part but I can't find a solution on how to extract the blocks.
dput(DF)
structure(list(VariableA = c(-0.427796831, -0.985783635, 0.07381913,
-0.788768923, 2.088999368, 1.634064399, -0.396180684, 1.242763624,
-0.925287904, -1.127545153, -1.392674655, -0.988900906, -0.08007986,
1.123984722, 0.698530819, -0.983565282, 0.568517376, -0.349446274,
0.451443794, -0.525897224, -0.932426185, -1.026114049, -0.502973503,
0.779152951, -0.636137726, -0.488850226, 0.281389897, -0.058183652,
-0.490377469, 0.541441864, 0.101754052, -0.16701156, 0.830697787,
0.383672008, 0.376444634, 0.377695822, -0.167281753, 0.85629382,
0.213632586, -0.180474289, 1.008370316, -0.039110304, -0.498537412,
-2.804652051, -0.308652164, -0.57234963, 0.599951896, 0.52484456,
0.008141731, -0.355182154, -0.401441593, 1.201478908, 0.656311257,
0.459034655), VariableB = c(-0.599169932, -0.874625086, -0.879367189,
0.068133167, -0.800781757, -0.746429115, -0.231178499, -0.905456972,
0.40165965, 0.664579078, -0.386614574, -0.700272577, 1.844891234,
0.277616227, 0.560119708, -2.874313318, 0.835592571, -0.66310824,
0.770336487, 1.547635124, -0.604065751, 1.009519877, -0.54792181,
-0.904229067, -0.309270319, 0.16088111, 0.325712725, -0.931632811,
-1.124531146, -0.24012375, -0.887921437, -1.531276383, 1.565233292,
0.462452663, 0.836271408, -0.721959208, 1.92215585, 0.189964832,
1.661140854, -1.604886269, -1.237132008, 0.811584528, -0.965798536,
2.604504203, -1.124331258, 0.240004185, -0.34902354, -0.447056073,
0.051475583, 0.159486311, -1.86620661, -1.671688795, -1.268626575,
-1.734731137), VariableC = structure(c(11L, 19L, 9L, 36L, 36L,
26L, 7L, 24L, 36L, 5L, 17L, 15L, 33L, 30L, 29L, 21L, 31L, 10L,
36L, 36L, 36L, 36L, 36L, 36L, 36L, 36L, 36L, 36L, 36L, 36L, 8L,
16L, 35L, 25L, 28L, 4L, 32L, 27L, 34L, 18L, 36L, 36L, 14L, 2L,
13L, 3L, 36L, 23L, 22L, 1L, 20L, 6L, 36L, 12L), .Label = c("-0.019569584",
"-0.020014785", "-0.033234545", "-0.034426339", "-0.046296608",
"-0.047020989", "-0.062735918", "-0.078616739", "-0.080554806",
"-0.101255451", "-0.102696676", "-0.127569648", "-0.143298342",
"-0.146433595", "-0.168917348", "-0.169828794", "-0.177928923",
"-0.178536056", "-0.186040872", "-0.22676482", "-0.38578786",
"0.005961731", "0.007778849", "0.033730665", "0.084612467", "0.088763528",
"0.104625865", "0.121271604", "0.125865053", "0.140160095", "0.140410995",
"0.17548741", "0.176481137", "0.187477344", "0.239593108", "FALSE"
), class = "factor")), .Names = c("VariableA", "VariableB", "VariableC"
), class = "data.frame", row.names = c(NA, -54L))

Answer 1

Here's an approach:
# create indicator variable
df$ind <- cumsum(df$VariableC == "FALSE")
# remove "FALSE" rows
df_sub <- df[df$VariableC != "FALSE", ]
# run a regression for each unique ind value
library(MASS)
lmList(VariableA ~ VariableB | ind, data = df_sub)
The result:
Call: lmList(formula = VariableA ~ VariableB | ind, data = df_sub)
Coefficients:
(Intercept) VariableB
0 -0.40531670 0.05261483
2 -0.93213791 -2.80237922
3 -0.26593782 0.31197216
15 0.24240710 0.10646927
17 -0.92256481 -0.65475348
18 0.02793152 -0.22209490
19 0.45903466 NA
Degrees of freedom: 35 total; 21 residual
Residual standard error: 0.6656342
How to create a plot?
library(ggplot2)
ggplot(df_sub, aes(x = VariableB, y = VariableA)) +
geom_point() +
facet_wrap( ~ ind) +
geom_smooth(method = lm)

Answer 2

You could do as follows:
falseIdx <- which(as.character(DF$VariableC) == "FALSE")
# at least 2 FALSE's must be present...
if(length(falseIdx) >= 2){
blocks <-
lapply(2:(length(falseIdx)-1),FUN=function(idx){
currFalse <- falseIdx[idx]
prevFalse <- falseIdx[idx-1]
# we build a block only if it has at least 10 rows
if(currFalse - prevFalse - 1 >= 10){
return(DF[(prevFalse+1):(currFalse-1),])
}else{
return(NULL)
}
})
# remove nulls
blocks[sapply(blocks, is.null)] <- NULL
}else{
blocks <- list()
}
Computing on your example data, blocks contains only one data.frame:
> blocks
[[1]]
VariableA VariableB VariableC
31 0.1017541 -0.8879214 -0.078616739
32 -0.1670116 -1.5312764 -0.169828794
33 0.8306978 1.5652333 0.239593108
34 0.3836720 0.4624527 0.084612467
35 0.3764446 0.8362714 0.121271604
36 0.3776958 -0.7219592 -0.034426339
37 -0.1672818 1.9221558 0.17548741
38 0.8562938 0.1899648 0.104625865
39 0.2136326 1.6611409 0.187477344
40 -0.1804743 -1.6048863 -0.178536056