Forvalues error: "i not found"

[Yingyi Lin]

Dear Statalist,

I have encountered a seemly simple question when I was trying to "forvalues" in STATA 15.

Before the forvalues, the codes were: (which works without any error)

generate new_var=old_var + 1 if inrange(_n,(1-1)*5760+1,5760*1)
replace   new_var=old_var + 2 if inrange(_n,(2-1)*5760+1,5760*2)
replace   new_var=old_var + 3 if inrange(_n,(3-1)*5760+1,5760*3)
replace   new_var=old_var + 4 if inrange(_n,(4-1)*5760+1,5760*4)
replace   new_var=old_var + 5 if inrange(_n,(5-1)*5760+1,5760*5)
replace   new_var=old_var + 6 if inrange(_n,(6-1)*5760+1,5760*6)
replace   new_var=old_var + 7 if inrange(_n,(7-1)*5760+1,5760*7)

Then, I tried to use "forvalues" to simplify my codes:

forvalues i=1(1)7{
gen neww_var=old_var + i if inrange(new_n,(i-1)*5760+1,5760*i)
format loop_date  %tdNN/DD/CCYY
}

And STATA log showed that:
i not found
r(111);

Anyone has a thought about that? Really don't know what is going wrong here...though it really seems a simple loop for me
Any reply would be very much appreciated! Thanks in advance!

Yingyi

[Marcos Almeida]

You should take a look at the Stata Manual. In short, you’ll need to use ‘i’ (typed in my portable phone, hence not ‘exactly’, but I believe you got the idea.

[Weiwen Ng]

... you’ll need to use ‘i’ ...

He means:

`i'

[Nick Cox]

Marcos and Weiwen are right. You must refer to a local macro that is a loop index with appropriate punctuation.

Inside your loop, it's not illegal to refer to i but it can only mean a variable or scalar. Stata is telling you that there is no such variable or scalar.

Even with that fixed your loop would fail as second time around you are trying to generate a variable which already exists.

That is all assuming that you have variables newn and loop_date (not obviously relevant here).

If you have loop_date, there is no point to assigning the same format seven times over. That statement belongs outside the loop.

All that said, you don't need a loop for what you want. A single command will suffice.

. clear

. set obs `=5760*7'
number of observations (_N) was 0, now 40,320

. gen oldvar = 42

. gen newvar = oldvar + ceil(7 * _n/_N)

. tab newvar

     newvar |      Freq.     Percent        Cum.
------------+-----------------------------------
         43 |      5,760       14.29       14.29
         44 |      5,760       14.29       28.57
         45 |      5,760       14.29       42.86
         46 |      5,760       14.29       57.14
         47 |      5,760       14.29       71.43
         48 |      5,760       14.29       85.71
         49 |      5,760       14.29      100.00
------------+-----------------------------------
      Total |     40,320      100.00

See help foreach and Statalist FAQ Advice #18

[Yingyi Lin]

Thanks so much for the reply guys! I ended up using Nick's codes and it works perfectly! Thanks again~

[Yingyi Lin]

Marcos and Weiwen are right. You must refer to a local macro that is a loop index with appropriate punctuation.

Inside your loop, it's not illegal to refer to i but it can only mean a variable or scalar. Stata is telling you that there is no such variable or scalar.

Even with that fixed your loop would fail as second time around you are trying to generate a variable which already exists.

That is all assuming that you have variables newn and loop_date (not obviously relevant here).

If you have loop_date, there is no point to assigning the same format seven times over. That statement belongs outside the loop.

All that said, you don't need a loop for what you want. A single command will suffice.

. clear

. set obs `=5760*7'
number of observations (_N) was 0, now 40,320

. gen oldvar = 42

. gen newvar = oldvar + ceil(7 * _n/_N)

. tab newvar

newvar | Freq. Percent Cum.
------------+-----------------------------------
43 | 5,760 14.29 14.29
44 | 5,760 14.29 28.57
45 | 5,760 14.29 42.86
46 | 5,760 14.29 57.14
47 | 5,760 14.29 71.43
48 | 5,760 14.29 85.71
49 | 5,760 14.29 100.00
------------+-----------------------------------
Total | 40,320 100.00

See help foreach and Statalist FAQ Advice #18

Hi Nick, you are so right that even after I use `i' in below , the codes did not work.

forvalues i=1(1)7{
gen loop_new_var = old_var + `i' if inrange( _n ,(`i'-1)*5760+1,5760*`i')
}

STATA logs showed: variable loop_new_var already defined

[Yingyi Lin]

Codes I ended up using:

gen loop_new_var
forvalues i=1(1)7{
replace loop_new_var = old_var + `i' if inrange( _n,(`i'-1)*5760+1,5760*`i')
}

OR

set obs `=5760*7'old_var
*number of observations (_N) was 0, now 40,320
gen new_var = old_var + ceil(7 * _n/_N)

Thanks Marcos, Weiwen and Nick, again!