Non numeric Characters in data set

[Yogesh Malhotra]

I am facing a problem in destring command while running it as my data set contains X00,X99,X10 values and other values are numeric. I am getting following result
destring nco2004, replace
nco2004 contains nonnumeric characters; no replace
Can you help sir with this problem? nco2004 is variable name

[Nick Cox]

Using the term "sir" is not a good idea: this is not a club for males only.

So, you need to find the non-numeric characters that are problematic.

tab nco2004 if missing(real(nco2004))

and then and only then can there be a diagnosis of

1. whether destring is the right command any way

2. if it is, which options to use (see help destring)

For more see e.g. http://www.statalist.org/forums/foru...iable-problems

[Yogesh Malhotra]

Thank you , I get it and got the result.

[Marcos Almeida]

This is a similar thread :http://www.statalist.org/forums/foru...meric-variable

There, some options to tackle the problem were commented, such as "force" and "ignore".

But the reason I share it here is that Nick's command in #2 provides a (much) more elegant solution.

[Omar Zahaf]

I am having a similar issue. I have a variable label called LeftKneeAlignment in a dataset called Joints. The variable label is currently a string variable but its meant to be all numerical values ranging from -16 to +23. In the variable list there are values that are non-numeric "N", "P", and "Z" which I am trying to ignore. The data is below:

Freq. Percent Cum.
------------+-----------------------------------
-1 | 372 12.29 12.29
-10 | 31 1.02 13.32
-11 | 14 0.46 13.78
-12 | 7 0.23 14.01
-13 | 4 0.13 14.14
-14 | 5 0.17 14.31
-15 | 9 0.30 14.61
-16 | 4 0.13 14.74
-18 | 4 0.13 14.87
-2 | 342 11.30 26.17
-20 | 1 0.03 26.21
-21 | 1 0.03 26.24
-3 | 284 9.39 35.62
-4 | 244 8.06 43.69
-5 | 155 5.12 48.81
-6 | 132 4.36 53.17
-7 | 77 2.54 55.72
-8 | 63 2.08 57.80
-9 | 40 1.32 59.12
0 | 363 12.00 71.12
1 | 259 8.56 79.68
10 | 5 0.17 79.84
11 | 3 0.10 79.94
12 | 3 0.10 80.04
14 | 1 0.03 80.07
15 | 1 0.03 80.11
2 | 209 6.91 87.01
3 | 131 4.33 91.34
4 | 59 1.95 93.29
5 | 41 1.35 94.65
6 | 27 0.89 95.54
7 | 15 0.50 96.03
8 | 9 0.30 96.33
9 | 7 0.23 96.56
N | 41 1.35 97.92
P | 10 0.33 98.25
Z | 53 1.75 100.00
------------+-----------------------------------
Total | 3,026 100.00

I am trying to categorise the dataset into 3 categories.

so <-2 = varus
-2 to +2 = neutral
>+2 = valgus

I have tried to ignore the "P,N and Z" and then categorise the remaining dataset by doing the following:

recode LeftKneeAlignment (min/-2=Varus) (-2/2=Neutral) (2/max=Valgus) ignore "N" "P" "Z"

But it keeps saying recode only allows numeric variables

Grateful for any help

[Rich Goldstein]

the "ignore" option goes with the -destring- command; see

help destring

after that you can use -recode- but you are not using that correctly so also see

help recode