Character to numeric: Error: NOTE: Invalid argument to function INPUT at line

[LisaYIN9309]

Hello,

I saw there are a few post about input statement errors but so far I haven't find one that solves my problem, so I am posting it here. I am doing a very basic character to numeric convert for a variable that supposed to be date value. My input data was import from CSV file.

Proc contents on my dataset after import as below. the variable I want to convert is

Alphabetic List of Variables and Attributes# Variable Type Len Format Informat35421

ENR_CODE	Num	8	BEST12.	BEST32.
ENR_END_DATE	Char	10	$10.	$10.
ENR_START_DATE	Num	8	MMDDYY10.	MMDDYY10.
PAYER_PLAN_ID	Char	7	$7.	$7.
PERSON_ID	Num	8	BEST12.	BEST32.

The missing values in this variables are entered as 'NI', so my first intuition is that this true character sting is preventing the conversion, so changed 'NI' to 0 then I did a simple input statement on this variable:

data test;

set pnt.oha_enro;

if ENR_END_DATE = 'NI' then ENR_end_date = '0';

run;

data test2;

set test;

try = input(ENR_END_DATE,8.0);

run;

Then I got this in the log window

data test2;

set test;

try = input(ENR_END_DATE,8.0);

run;

NOTE: Invalid argument to function INPUT at line 90 column 7.

Anyone has an idea why it's happening?

Thank you so much

[Reeza]

What does that variable look like? And do you have missing values? Missing would generate that error as well.

---

@LisaYIN9309 wrote:

Hello,

I saw there are a few post about input statement errors but so far I haven't find one that solves my problem, so I am posting it here. I am doing a very basic character to numeric convert for a variable that supposed to be date value. My input data was import from CSV file.

Proc contents on my dataset after import as below. the variable I want to convert is

Alphabetic List of Variables and Attributes# Variable Type Len Format Informat35421

ENR_CODE	Num	8	BEST12.	BEST32.
ENR_END_DATE	Char	10	$10.	$10.
ENR_START_DATE	Num	8	MMDDYY10.	MMDDYY10.
PAYER_PLAN_ID	Char	7	$7.	$7.
PERSON_ID	Num	8	BEST12.	BEST32.

The missing values in this variables are entered as 'NI', so my first intuition is that this true character sting is preventing the conversion, so changed 'NI' to 0 then I did a simple input statement on this variable:

data test;

set pnt.oha_enro;

if ENR_END_DATE = 'NI' then ENR_end_date = '0';

run;

data test2;

set test;

try = input(ENR_END_DATE,8.0);

run;

Then I got this in the log window

data test2;

set test;

try = input(ENR_END_DATE,8.0);

run;

NOTE: Invalid argument to function INPUT at line 90 column 7.

Anyone has an idea why it's happening?

Thank you so much

---

[ballardw · Accepted solution]

I am going to guess that the non "NI" values of ENR_END_DATE actually look more like the MMDDYY10. from ENR_START_DATE.

In which case you would want to 1) use the mmddyy10 informat, 2) instead of '0' assign . for missing 3) assign an appropriate format such as mmddyy10. for the new variable.

Or go back to the import step, copy the data step code created by proc import (assumed) from the log, paste into the editor and change the informat and format to mmyydd10. perhaps add ?? after the variable name in the input statement to suppress invalid date messages for that variable reading the data.

[LisaYIN9309]

Thanks @ballardw, I changed the 'NI' to a mmddyy10. format value instead of 0 and it worked! seems like 0 is the problem. There is no other missing values.

Thank you again

[Kurt_Bremser]

The most elegant method to do this is

try = ifn(enr_end_date = 'NI',.,input(enr_end_date,mmddyy10.));

[Kurt_Bremser]

A 10-character variable with "date" in the name lets me think it will contain values like 09/04/2018, which can't be input with an 8. informat. If the values are like that, use a proper date informat.