CODEWITHSUNDEEP
r
user1100825
user1100825

Reputation: 117

Find first non-zero occurrence in dataframe

I have a time-series of Sales by Account ID. To calculate average growth, I need to extract the first month with non-zero sales for each ID. Since the account could have been established at different times, I need to dynamically identify when sales > 0 for the first time in the account.

The index to the row would be sufficient for me to pass to a function calculating growth. So I expect the following results by Account ID:

54 - [1]
87 - [4]
95 - [2]

I tried `apply(df$Sales,2,match,x>0)`  but this doesn't work.

Any pointers? Alternatively, is there an easier way to compute CAGR with this dataset?

Thanks in advance!

CalendarMonth   ID  Sales
8/1/2008    54  6692.60274
9/1/2008    54  6476.712329
10/1/2008   54  6692.60274
11/1/2008   54  6476.712329
12/1/2008   54  11098.60822
7/1/2008    87  0
8/1/2008    87  0
9/1/2008    87  0
10/1/2008   87  18617.94155
11/1/2008   87  18017.36279
12/1/2008   87  18617.94155
1/1/2009    87  18617.94155
2/1/2009    87  16816.20527
7/1/2008    95  0
8/1/2008    95  8015.956284
9/1/2008    95  0
10/1/2008   95  8015.956284
11/1/2008   95  6309.447514
12/1/2008   95  6519.762431
1/1/2009    95  6519.762431

Upvotes: 6

Views: 7427

Answers (3)

mbask
mbask

Reputation: 2481

Building up on digEmAll answer, a solution using functional programming (maybe a bit cleaner):

> res3 <- tapply(
  1:nrow(df)
  , df$ID
  , function(Idx) Idx[Position(function(x) df[x, "Sales"] > 0, Idx)]
)
> identical(res3, res2)
[1] TRUE

Upvotes: 1

A_K
A_K

Reputation: 2651

Would this help:

tapply(df$Sales, df$ID, function(a)head(which(a>0),1))

where df is your data frame above?

If you want the entire row & not just the index, this might help:

lapply(unique(df$ID),function(a) head(subset(df,ID==a & Sales>0),1))

Upvotes: 9

digEmAll
digEmAll

Reputation: 57220

Here's a possible solution:

res1 <- tapply(df$Sales,INDEX=df$ID,FUN=function(x) which(x > 0)[1])

> res1
54 87 95 
 1  4  2

Where res is a numeric vector with :

> names(res)
[1] "54" "87" "95"

If you want to get the indexes of the row in the original data.frame and not in the subsets, you can do:

res2 <- tapply(1:nrow(df),
              INDEX=df$ID,FUN=function(idxs) idxs[df[idxs,'Sales'] > 0][1])

> res2
54 87 95 
 1  9 15

Then you can simply use the indexes in res2, to subset the data.frame:

df2 <- df[res2,]

> df2 
CalendarMonth   ID      Sales
  8/1/2008      54     6692.603
 10/1/2008      87    18617.942
  8/1/2008      95     8015.956

Upvotes: 3

Related Questions