Mean of quartile for multiple columns and multiple dates - r

I'm trying to find the mean forward return (column fwd_rtn) of each quartile for each column (ie for quartiles for PB, PE, PS) for each date group (1/1/2016...1/4/2016)
head(df)
Date Stock Price PB PE PS fwd_rtn
1 1/1/2016 A 11.90 0.4 0.10 0.57 -0.015
2 1/1/2016 B 3.56 0.8 0.09 0.26 -0.036
3 1/1/2016 C 1.29 1.2 0.18 1.60 0.10
......
4 1/4/2016 A 12.80 0.39 0.13 0.53 -0.01
5 1/4/2016 B 4.03 0.76 0.08 0.23 0.02
6 1/4/2016 C 1.83 0.87 0.14 1.16 0.03
So far i have been able to find the mean return for 1 column for 1 date using this code
df$qPB <- cut(df$PB, breaks = quantile(df$PB, c(0,.25,.5,.75,1)),include.lowest = TRUE)
aggregate(df$fwd_rtn,list(qPB = df$qPB),FUN=mean)
which gave me the right answers. But I'm struggling to do it for the multiple columns. I think I'm supposed to use dplyr and the gather() function but i dont know how.

To get quartiles of a single variable by date you can use the ave function:
df$qPB <- ave(df$PB, df$Date, FUN= function(i) cut(i, breaks = quantile(df$PB,
c(0,.25,.5,.75,1)),include.lowest = TRUE)
# a minor addition to aggregate
aggregate(df$fwd_rtn, list("qPB"=df$qPB, "date"=df$Date), FUN=mean)
You should take a look at using lapply or sapply to move through multiple columns.

Related

how to use the `map` family command in **purrr** pacakge to swap the columns across rows in data frame?

Imagine there are 4 cards on the desk and there are several rows of them (e.g., 5 rows in the demo). The value of each card is already listed in the demo data frame. However, the exact position of the card is indexed by the pos columns, see the demo data I generated below.
To achieve this, I swap the cards with the [] function across the rows to switch the cards' values back to their original position. The following code already fulfills such a purpose. To avoid explicit usage of the loop, I wonder whether I can achieve a similar effect if I use the vectorization function with packages from tidyverse family, e.g. pmap or related function within the package purrr?
# 1. data generation ------------------------------------------------------
rm(list=ls())
vect<-matrix(round(runif(20),2),nrow=5)
colnames(vect)<-paste0('card',1:4)
order<-rbind(c(2,3,4,1),c(3,4,1,2),c(1,2,3,4),c(4,3,2,1),c(3,4,2,1))
colnames(order)=paste0('pos',1:4)
dat<-data.frame(vect,order,stringsAsFactors = F)
# 2. data swap ------------------------------------------------------------
for (i in 1:dim(dat)[1]){
orders=dat[i,paste0('pos',1:4)]
card=dat[i,paste0('card',1:4)]
vec<-card[order(unlist(orders))]
names(vec)=paste0('deck',1:4)
dat[i,paste0('deck',1:4)]<-vec
}
dat
You could use pmap_dfr :
card_cols <- grep('card', names(dat))
pos_cols <- grep('pos', names(dat))
dat[paste0('deck', seq_along(card_cols))] <- purrr::pmap_dfr(dat, ~{
x <- c(...)
as.data.frame(t(unname(x[card_cols][order(x[pos_cols])])))
})
dat
# card1 card2 card3 card4 pos1 pos2 pos3 pos4 deck1 deck2 deck3 deck4
#1 0.05 0.07 0.16 0.86 2 3 4 1 0.86 0.05 0.07 0.16
#2 0.20 0.98 0.79 0.72 3 4 1 2 0.79 0.72 0.20 0.98
#3 0.50 0.79 0.72 0.10 1 2 3 4 0.50 0.79 0.72 0.10
#4 0.03 0.98 0.48 0.06 4 3 2 1 0.06 0.48 0.98 0.03
#5 0.41 0.72 0.91 0.84 3 4 2 1 0.84 0.91 0.41 0.72
One thing to note here is to make sure that the output from pmap function does not have original names of the columns. If they have the original names, it would reshuffle the columns according to the names and output would not be in correct order. I use unname here to remove the names.

How can I get row-wise max based on condition of specific column in R dataframe?

I'm trying to get the maximum value BY ROW across several columns (climatic water deficit -- def_59_z_#) depending on how much time has passed (time since fire -- YEAR.DIFF). Here are the conditions:
If 1 year has passed, select the deficit value for first year.
(def_59_z_1).
If 2 years: max deficit of first 2 years.
If 3 years: max of deficit of first 3 years.
If 4 years: max of deficit of first 4 years.
If 5 or more years: max of first 5 years.
However, I am unable to extract a row-wise max when I include a condition. There are several existing posts that address row-wise min and max (examples 1 and 2) and sd (example 3) -- but these don't use conditions. I've tried using apply but I haven't been able to find a solution when I have multiple columns involved as well as a conditional requirement.
The following code simply returns 3.5 in the new column def59_z_max15, which is the maximum value that occurs in the dataframe -- except when YEAR.DIFF is 1, in which case def_50_z_1 is directly returned. But for all the other conditions, I want 0.98, 0.67, 0.7, 1.55, 1.28 -- values that reflect the row maximum of the specified columns. Link to sample data here. How can I achieve this?
I appreciate any/all suggestions!
data <- data %>%
mutate(def59_z_max15 = ifelse(YEAR.DIFF == 1,
(def59_z_1),
ifelse(YEAR.DIFF == 2,
max(def59_z_1, def59_z_2),
ifelse(YEAR.DIFF == 3,
max(def59_z_1, def59_z_2, def59_z_3),
ifelse(YEAR.DIFF == 4,
max(def59_z_1, def59_z_2, def59_z_3, def59_z_4),
max(def59_z_1, def59_z_2, def59_z_3, def59_z_4, def59_z_5))))))
Throw this function in an apply family function
func <- function(x) {
first.val <- x[1]
if (first.val < 5) {
return(max(x[2:(first.val+)])
} else {
return(max(x[2:6]))
}
}
Your desired output should be obtained by:
apply(data, 1, function(x) func(x)) #do it by row by setting arg2 = 1
An option would be to get the pmax (rowwise max - vectorized) for each set of conditions separately in a loop (map - if the value of 'YEAR.DIFF' is 1, select only the 'def_59_z_1', for 2, get the max of 'def_59_z_1' and 'def_59_z_2', ..., for 5, max of 'def_59_z_1' to 'def_59_z_5', coalesce the columns together and replace the rest of the NA with the pmax of all the 'def59_z" columns
library(tidyverse)
out <- map_dfc(1:5, ~
df1 %>%
select(seq_len(.x) + 1) %>%
transmute(val = na_if((df1[["YEAR.DIFF"]] == .x)*
pmax(!!! rlang::syms(names(.))), 0))) %>%
transmute(def59_z_max15 = coalesce(!!! rlang::syms(names(.)))) %>%
bind_cols(df1, .)%>%
mutate(def59_z_max15 = case_when(is.na(def59_z_max15) ~
pmax(!!! rlang::syms(names(.)[2:6])), TRUE ~ def59_z_max15))
head(out, 10)
# YEAR.DIFF def59_z_1 def59_z_2 def59_z_3 def59_z_4 def59_z_5 def59_z_max15
#1 5 0.25 -2.11 0.98 -0.07 0.31 0.98
#2 9 0.67 0.65 -0.27 0.52 0.26 0.67
#3 10 0.56 0.33 0.03 0.70 -0.09 0.70
#4 2 -0.34 1.55 -1.11 -0.40 0.94 1.55
#5 4 0.98 0.71 0.41 1.28 -0.14 1.28
#6 3 0.71 -0.17 1.70 -0.57 0.43 1.70
#7 4 -1.39 -1.71 -0.89 0.78 1.22 0.78
#8 4 -1.14 -1.46 -0.72 0.74 1.32 0.74
#9 2 0.71 1.39 1.07 0.65 0.29 1.39
#10 1 0.28 0.82 -0.64 0.45 0.64 0.28
data
df1 <- read.csv("https://raw.githubusercontent.com/CaitLittlef/random/master/data.csv")

Column Mean for rows with unique values

how can I compute the mean R, R1, R2, R3 values from the rows sharing the same lon,lat field? I'm sure this questions exists multiple times but I could not easily find it.
lon lat length depth R R1 R2 R3
1 147.5348 -35.32395 13709 1 0.67 0.80 0.84 0.83
2 147.5348 -35.32395 13709 2 0.47 0.48 0.56 0.54
3 147.5348 -35.32395 13709 3 0.43 0.29 0.36 0.34
4 147.4290 -35.27202 12652 1 0.46 0.61 0.60 0.58
5 147.4290 -35.27202 12652 2 0.73 0.96 0.95 0.95
6 147.4290 -35.27202 12652 3 0.77 0.92 0.92 0.91
I'd recommend using the split-apply-combine strategy, where you're splitting by BOTH lon and lat, applying mean to each group, then recombining into a single data frame.
I'd recommend using dplyr:
library(dplyr)
mydata %>%
group_by(lon, lat) %>%
summarize(
mean_r = mean(R)
, mean_r1 = mean(R1)
, mean_r2 = mean(R2)
, mean_r3 = mean(R3)
)

apply a function on columns with specific names

I am new in R.
I have hundreds of data frames like this
ID NAME Ratio_A Ratio_B Ratio_C Ratio_D
AA ABCD 0.09 0.67 0.10 0.14
AB ABCE 0.04 0.85 0.04 0.06
AC ABCG 0.43 0.21 0.54 0.14
AD ABCF 0.16 0.62 0.25 0.97
AF ABCJ 0.59 0.37 0.66 0.07
This is just an example. The number and names of the Ratio_ columns are different between data frames, but all of them start with Ratio_. I want to apply a function (for example, log(x)), to the Ratio_ columns without specify the column number or the whole name.
I know how to do it df by df, for the one in the example:
A <- function(x) log(x)
df_log<-data.frame(df[1:2], lapply(df[3:6], A))
but I have a lot of them, and as I said the number of columns is different in each.
Any suggestion?
Thanks
Place the datasets in a list and then loop over the list elements
lapply(lst, function(x) {i1 <- grep("^Ratio_", names(x));
x[i1] <- lapply(x[i1], A)
x})
NOTE: No external packages are used.
data
lst <- mget(paste0("df", 1:100))
This type of problem is very easily dealt with using the dplyr package. For example,
df <- read.table(text = 'ID NAME Ratio_A Ratio_B Ratio_C Ratio_D
AA ABCD 0.09 0.67 0.10 0.14
AB ABCE 0.04 0.85 0.04 0.06
AC ABCG 0.43 0.21 0.54 0.14
AD ABCF 0.16 0.62 0.25 0.97
AF ABCJ 0.59 0.37 0.66 0.07',
header = TRUE)
library(dplyr)
df_transformed <- mutate_each(df, funs(log(.)), starts_with("Ratio_"))
df_transformed
# > df_transformed
# ID NAME Ratio_A Ratio_B Ratio_C Ratio_D
# 1 AA ABCD -2.4079456 -0.4004776 -2.3025851 -1.96611286
# 2 AB ABCE -3.2188758 -0.1625189 -3.2188758 -2.81341072
# 3 AC ABCG -0.8439701 -1.5606477 -0.6161861 -1.96611286
# 4 AD ABCF -1.8325815 -0.4780358 -1.3862944 -0.03045921
# 5 AF ABCJ -0.5276327 -0.9942523 -0.4155154 -2.65926004

Ordering Table A based on Rank of Table B in R

pretty newb question here, but I have not been able to track down a solution for some time:
I have an XTS object of trading indicators (indicate) for stock data that looks like
A XOM MSFT
2000-11-30 -0.59 0.22 0.10
2000-12-29 0.55 -0.23 0.05
2001-01-30 -0.52 0.09 -0.10
And a table with an identical index for the corresponding period returns (return) that looks like
A XOM MSFT
2000-11-30 -0.15 0.10 0.03
2000-12-29 0.03 -0.05 0.02
2001-01-30 -0.04 0.02 -0.05
I have sorted the indicator table and had it return the column name with the following code:
indicate.label <- colnames(indicate)
indicate.rank <- t(apply(indicate, 1, function(x) indicate.label[order(-x)]))
indicate.rank <- xts(indicate.rank, order.by = index(returns))
Which gives the table (indicate.rank) of the symbol names ranked by their trading indicator:
1 2 3
2000-11-30 XOM MSFT A
2000-12-29 A MSFT XOM
2001-01-30 XOM A MSFT
I would like to also have a table that gives the period returns based on the indicator rank:
2000-11-30 0.10 0.03 -0.15
2000-12-29 0.03 0.02 -0.05
2001-01-30 0.02 -0.04 -0.05
I cannot figure out how to call the correct symbol for all rows or just sort the table return based on the order of indicate.
Thank you for any suggestions.
Trevor J
I'm not particularly satisfied with this solution, but it works.
row.rank <- t(apply(indicate, 1, order, decreasing=TRUE))
indicate.rank <- return.rank <- indicate # pre-allocate
for(i in 1:NROW(indicate.rank)) {
indicate.rank[i,] <- colnames(indicate)[row.rank[i,]]
return.rank[i,] <- return[i,row.rank[i,]]
}
It would probably be easier to handle this if the returns and the indicators for each symbol were in the same object, but I don't know how that would fit with the rest of your workflow.

Resources