Troubles with handling vector data (csv format) in R
2
1
Entering edit mode
6 weeks ago

I am trying to convert vector data in csv format from this form 1 to this form 2 and vice versa in R. This has been giving me sleepless nights. I have been doing this in wolfram in the past but I had to switch to R. I would be very grateful for your kind suggestions. Thanks

csv data R frame • 519 views
1
Entering edit mode

what is represented in columns B and C of the desired output?

0
Entering edit mode

rpolicastro, they are actually cell identifiers in MS Excel. In reality, the first column represents the field of view in x-direction, second column is field of view in y-direction while the third column represents the intensity distribution of an image.

2
Entering edit mode

Looks like wide to long form. Check melt() from the reshape2 package or pivot_longer from tidyr.

3
Entering edit mode
6 weeks ago

Example data.

df <- as.data.frame(cbind(seq_len(10), replicate(5, rnorm(10, 10, 2))))
colnames(df) <- seq_len(ncol(df))

> df
1         2         3         4         5         6
1   1 12.702893  8.850859  7.680388  7.490946  9.457605
2   2 10.770156 10.632835  8.534555  8.273824  8.423220
3   3  6.399747 11.745845  7.283831 14.609756 12.841378
4   4  9.832645  7.585846  8.680142  5.917577 10.592092
5   5 14.073382 11.741811  9.235911  9.279248 14.884535
6   6  8.802907 10.402209  8.652651 11.458196  7.249231
7   7 12.076647 12.979112 10.461552 13.131768 13.275357
8   8 11.231053  9.862153 11.059780  8.599320 12.705250
9   9  6.318269 13.033082 10.937295  9.703796  9.578383
10 10 13.427937  8.932672 10.892284 10.935992 13.070986


library("tidyr")

df_long <- pivot_longer(df, !1)

> df_long
# A tibble: 50 × 3
1 name  value
<dbl> <chr> <dbl>
1     1 2     12.7
2     1 3      8.85
3     1 4      7.68
4     1 5      7.49
5     1 6      9.46
6     2 2     10.8
7     2 3     10.6
8     2 4      8.53
9     2 5      8.27
10     2 6      8.42
# … with 40 more rows


data.table answer (faster than tidyverse with a lot of data)

library("data.table")

setDT(df)
df_long <- melt(df, id.vars=1)

> df_long[1:10, ]
1 variable     value
1:  1        2 12.702893
2:  2        2 10.770156
3:  3        2  6.399747
4:  4        2  9.832645
5:  5        2 14.073382
6:  6        2  8.802907
7:  7        2 12.076647
8:  8        2 11.231053
9:  9        2  6.318269
10: 10        2 13.427937

2
Entering edit mode
6 weeks ago
Hamid Ghaedi ★ 2.0k

For the sake of completeness, I am adding this reply to add another method to what rpolicastro posted and considering ATpoint 's comment :

This is wide to long-form conversion and vice versa scenario. My understanding from the post is the first two columns are x-y coordinates (id variables) and the rest are image intensities at the corresponding coordinates (measure variables). You want to convert from wide to long-form and long-form to wide:

df <- as.data.frame(cbind(seq_len(10), rep(0, 10), replicate(5, rnorm(10, 10, 2))))
colnames(df) <- c("A", "B", "C", "D", "E", "F", "G")

library(reshape2)

# convert df from wide  to long form by melt() function
meltDF = melt(df, id.vars = c("A", "B"))

#   A B variable     value
# 1 1 0        C  7.499697
# 2 2 0        C 10.163781
# 3 3 0        C  8.862005
# 4 4 0        C  9.305449
# 5 5 0        C 10.820081
# 6 6 0        C 10.524560

# convert meltDF from long form back to wide form using dcast() function
castDF = dcast(meltDF, A+B ~ variable, value.var = "value")

#   A B         C         D         E         F         G
# 1 1 0  7.499697  6.492901 10.196289  8.304000 10.835184
# 2 2 0 10.163781 10.269959 11.465557  7.326901  8.174221
# 3 3 0  8.862005 15.052346 12.191936 11.098902  7.913761
# 4 4 0  9.305449  8.986819 12.696276  9.643824  9.675121
# 5 5 0 10.820081  8.908355  8.291135 10.148121  7.451699
# 6 6 0 10.524560  8.256329 10.242652 10.824888  9.351173

1
Entering edit mode

From what I gathered from OP It would be meltDF <- melt(df, id.vars ="A") and castDF <- dcast(meltDF, A ~ variable, value.var = "value"), since the column names represent the y coordinates.

As a side note reshape2 has retired by the author and folded into tidyr [source].