[R] Capturing positive and negative changes using R

[Daniel Nordlund]

Here is one more option using the ave() function. Using Jim's data and 
naming convention

fkdf$X1_change <- ave(fkdf[,'X1'], fkdf$Country, FUN=function(x) 
c(0,diff(x)))
fkdf$X2_change <- ave(fkdf[,'X2'], fkdf$Country, FUN=function(x) 
c(0,diff(x)))

hope this is helpful,

Dan

-- 
Daniel Nordlund
Port Townsend, WA  USA


On 7/20/2019 4:17 PM, Jim Lemon wrote:
> Hi Faradj,
> Rui's advice is correct, here's a way to do it. Note that I have
> replaced the comma decimal points with full stops for my convenience:
>
> fkdf<-read.csv(text="Year,Country,X1,X2
> 1990,United States,0,0.22
> 1991,United States,0,0.22
> 1992,United States,0,0.22
> 1993,United States,0,0.22
> 1994,United States,0,0.22
> 1995,United States,0,0.22
> 1996,United States,0,0.22
> 1997,United States,0,0.5
> 1998,United States,0,0.5
> 1999,United States,0,0.5
> 2000,United States,0,0.5
> 2001,United States,0,0.5
> 2002,United States,2,NA
> 2003,United States,2,0.5
> 2004,United States,2,1
> 2005,United States,1,1
> 2006,United States,1,1
> 2007,United States,1,1
> 2008,United States,1,1
> 2009,United States,1,1
> 2010,United States,1,0.5
> 2011,United States,0,0.5
> 1990,Canada,1,1.5
> 1991,Canada,1,1.5
> 1992,Canada,1,NA
> 1993,Canada,1,1.5
> 1994,Canada,1,1.5
> 1995,Canada,1,1.5
> 1996,Canada,1,1.5
> 1997,Canada,1,1.5
> 1998,Canada,1,2
> 1999,Canada,2,2
> 2000,Canada,2,2
> 2001,Canada,2,2
> 2002,Canada,2,2
> 2003,Canada,1,2
> 2004,Canada,2,0.5
> 2005,Canada,1,0.5
> 2006,Canada,0,0.5
> 2007,Canada,1,0.5
> 2008,Canada,0,0.5
> 2009,Canada,1,0.5
> 2010,Canada,1,0.5
> 2011,Canada,0,1",
> header=TRUE,stringsAsFactors=FALSE)
> diffX1<-aggregate(fkdf$X1,by=list(fkdf[,2]),FUN=diff)
> diffX2<-aggregate(fkdf$X2,by=list(fkdf[,2]),FUN=diff)
> diffX1<-data.frame(diffX1$Group.1,diffX1$x)
> diffyears<-unique(fkdf$Year)[-1]
> names(diffX1)<-c("Country",diffyears)
> diffX2<-data.frame(diffX2$Group.1,diffX2$x)
> names(diffX2)<-c("Country",diffyears)
>
> Jim
>
> On Sun, Jul 21, 2019 at 5:34 AM Faradj Koliev <faradj.g using gmail.com> wrote:
>> Dear R-users,
>>
>> I have a country-year data for 180 countries from 1970 to 2010. I’m interested in capturing positive and negative changes in some of the variables. Some of these variables are continuous (0,25, 0,33, 1, 1,5 etc) others are ordered (0,1, 2).
>>
>> To do this, I use this code data$X1_change<- +c(FALSE,diff(data$X1))
>>
>> My data looks something like this (please see below).
>>
>> There’re some problems with this code:  (1) I can’t capture the smaller changes, say from 0,25 to 0,33 ( I get weird numbers). I would love to get the exact difference ( for ex: +1, -0,22, +4, -2 etc).  (2) It can’t make difference between countries. That is, it takes the difference between countries while it should only do this for each country ( for ex: when the US ends in 2011, and Canada starts, it counts this a difference but it shouldn’t, see below). (3) NAs, missing values, is neither a positive or negative change, although it does think that what comes after the NA is a difference.
>>
>>   So, I wonder if anyone here can help me to adjust this code. I appreciate all comments.
>>
>>