Find differences betwen 2 dataframes with different lengths - r

Question

I have two dataframes with each two columns c("price", "size") with different lengths.
Each price must be linked to its size. It's two lists of trade orders. I have to discover the differences between the two dataframes knowing that the two databases can have orders that the other doesn't have and vice versa. I would like an output with the differences or two outputs, it doesn't matter. But I need the row number in the output to find where are the differences in the series.
Here is sample data :
> out
price size
1: 36024.86 0.01431022
2: 36272.00 0.00138692
3: 36272.00 0.00277305
4: 36292.57 0.05420000
5: 36292.07 0.00403948
---
923598: 35053.89 0.30904890
923599: 35072.76 0.00232000
923600: 35065.60 0.00273000
923601: 35049.36 0.01760000
923602: 35037.23 0.00100000
>bit
price size
1: 37279.89 0.01340020
2: 37250.84 0.00930000
3: 37250.32 0.44284049
4: 37240.00 0.00056491
5: 37215.03 0.99891906
---
923806: 35053.89 0.30904890
923807: 35072.76 0.00232000
923808: 35065.60 0.00273000
923809: 35049.36 0.01760000
923810: 35037.23 0.00100000
For example, I need to know if the first row of the database out is in the database bit.
I've tried many functions : comparedf()
summary(comparedf(bit, out, by = c("price","size"))
but I've got error:
Error in vecseq(f__, len__, if (allow.cartesian || notjoin ||
!anyDuplicated(f__, :
I've tried compare_df() :
compareout=compare_df(out,bit,c("price","size"))
But I know the results are wrong, I've only 23 results and I know that there are more than 200 differences minimum.
I've tried match(), which() functions but it doesn't get the results I search.
If you have any other methods, I will take them.

Answer 1

Perhaps you could just do inner_join on out and bit by price and size? But first make id variable for both data.frame's
library(dplyr)
out$id <- 1:nrow(out)
bit$id <- 1:nrow(bit)
joined <- inner_join(bit, out, by = c("price", "size"))
Now we can check which id from out and bit are not present in joined table:
id_from_bit_not_included_in_out <- bit$id[!bit$id %in% joined$id.x]
id_from_out_not_included_in_bit <- out$id[!out$id %in% joined$id.y]
And these ids are the rows not included in out or bit, i.e. variable id_from_bit_not_included_in_out contains rows present in bit, but not in out and variable id_from_out_not_included_in_bit contains rows present in out, but not in bit

Answer 2

First attempt here. It will be difficult to do a very clean job with this data tho.
The data I used:
out <- read.table(text = "price size
36024.86 0.01431022
36272.00 0.00138692
36272.00 0.00277305
36292.57 0.05420000
36292.07 0.00403948
35053.89 0.30904890
35072.76 0.00232000
35065.60 0.00273000
35049.36 0.01760000
35037.23 0.00100000", header = T)
bit <- read.table(text = "price size
37279.89 0.01340020
37250.84 0.00930000
37250.32 0.44284049
37240.00 0.00056491
37215.03 0.99891906
37240.00 0.00056491
37215.03 0.99891906
35053.89 0.30904890
35072.76 0.00232000
35065.60 0.00273000
35049.36 0.01760000
35037.23 0.00100000", header = T)
Assuming purely that row 1 of out should match with row 1 of bit a simple solution could be:
df <- cbind(distinct(out), distinct(bit))
names(df) <- make.unique(names(df))
However judging from the data you have provided I am not sure if this is the way to go (big differences in the first few rows) so maybe try sorting the data first?:
df <- cbind(distinct(out[order(out$price, out$size),]), distinct(bit[order(bit$price, bit$size),]))
names(df) <- make.unique(names(df))