# [R] back tick names with predict function

Robert Baer rb@er @end|ng |rom @t@u@edu
Fri Dec 1 01:04:39 CET 2023

```Thank you Rui.  I didn't know about the check.names = FALSE argument.
Another good reminder to always read help, but I'm not sure I understood
what help to read in this case.  Since your clue, I've discovered that a
tibble-based strategy could also work.

x = seq(min(cob_wt\$`plant-density`), max(cob_wt\$`plant-density`), length
= 999)
xvals = tibble(`plant-density` = x)
CI.c = predict(mod2, newdata = xvals, interval = 'c')
CI.p = predict(mod2,  newdata = xvals,  interval = 'p')

Again, appreciate the solution.

On 11/30/2023 12:03 PM, Rui Barradas wrote:
> Às 17:57 de 30/11/2023, Rui Barradas escreveu:
>> Às 17:38 de 30/11/2023, Robert Baer escreveu:
>>> I am having trouble using back ticks with the R extractor function
>>> 'predict' and an lm() model. I'm trying too construct some nice
>>> vectors that can be used for plotting the two types of regression
>>> intervals.  I think it works with normal column heading names but it
>>> fails when I have "special" back-tick names.  Can anyone help with
>>> how I would reference these?  Short of renaming my columns, is there
>>> a way to accomplish this?
>>>
>>> Repex
>>>
>>> *# dataframe with dashes in column headings
>>> cob =
>>>    structure(list(`cob-wt` = c(212, 241, 215, 225, 250, 241, 237,
>>>                              282, 206, 246, 194, 241, 196, 193, 224,
>>> 257, 200, 190, 208, 224
>>> ), `plant-density` = c(137, 107, 132, 135, 115, 103, 102, 65,
>>>                         149, 85, 173, 124, 157, 184, 112, 80, 165,
>>> 160, 157, 119)),
>>> class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, -20L))
>>>
>>> # regression model works
>>> mod2 = lm(`cob-wt` ~ `plant-density`, data = cob)
>>>
>>> # x sequence for plotting CI's
>>> # Set up x points
>>> x = seq(min(cob\$`plant-density`), max(cob\$`plant-density`), length =
>>> 1000)
>>>
>>> # Use predict to get CIs for a plot
>>> # Add CI for regression line (y-hat uses 'c')
>>> # usual trick is to assign x to actual x-var name in middle
>>> dataframe arguement
>>> CI.c = predict(mod2, data.frame( `plant-density` = x), interval =
>>> 'c') # fail
>>>
>>> # Add CI for prediction value (y-tilde uses 'p')
>>> # usual trick is to assign x to actual x-var name in middle
>>> dataframe arguement
>>> CI.p = predict(mod2, data.frame(`plant-density`  = x), interval =
>>> 'p')    # fail
>>> *
>>>
>>> ______________________________________________
>>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>> http://www.R-project.org/posting-guide.html
>>> and provide commented, minimal, self-contained, reproducible code.
>> Hello,
>>
>> When creating the new data df, the default check.names = TRUE changes
>> the column name, it is repaired and the hyphen is replaced by a legal
>> dot.
>>
>>
>> # check.names defaults to TRUE
>> newd <- data.frame(`plant-density` = x)
>> # `plant-density` is not a column name
>>
>> # check.names set to FALSE
>> newd <- data.frame(`plant-density` = x, check.names = FALSE)
>> # `plant-density` is becomes a column name
>>
>>
>> # Use predict to get CIs for a plot
>> # Add CI for regression line (y-hat uses 'c')
>> # usual trick is to assign x to actual x-var name in middle dataframe
>> arguement
>> CI.c = predict(mod2, newdata = newd, interval = 'confidence')  # fail
>>
>> # Add CI for prediction value (y-tilde uses 'p')
>> # usual trick is to assign x to actual x-var name in middle dataframe
>> arguement
>> CI.p = predict(mod2, newdata = newd, interval = 'prediction') # fail
>>
>>
>>
>> Hope this helps,
>>
>>
>>
> Hello,
>
> Sorry for the comments '# fail' in the last two instructions, I should
> have changed them.
>
>
> CI.c <- predict(mod2, newdata = newd, interval = 'confidence') # works
> CI.p <- predict(mod2, newdata = newd, interval = 'prediction') # works
>
>
> Hoep this helps,
>