How to extract p-values of regression stored as list

[amare]

Hi,
Your example worked out and but do you think I can save the result in CSV format?
Thanks!

[siddharthprabhu]

We could use bind_rows() to collapse the list of data frames into one like this:

library(tidyverse)
library(broom)

data <- data.frame(BMI = c(30, 23, 27, 35, 12, 10, 17, 24, 21, 15, 16),
                   g1 = c(35, 2, 3, 4, 5, 6, 7, 10, 12, 41, 76), 
                   g2 = c(20, 2, 7, 2, 8, 5, 5, 3, 7, 2, 12), 
                   g3 = c(5, 0, 4, 5, 2, 4, 8, 9, 20, 1, 11))

output <- data %>% 
  pivot_longer(-BMI) %>% 
  group_split(name) %>% 
  set_names(nm = map(., ~ first(.x$name))) %>% 
  map(~ tidy(lm(BMI ~ value, data = .))) %>% 
  map(~ filter(., p.value < 0.05)) %>% 
  bind_rows(.id = "var")

output
#> # A tibble: 3 x 6
#>   var   term        estimate std.error statistic   p.value
#>   <chr> <chr>          <dbl>     <dbl>     <dbl>     <dbl>
#> 1 g1    (Intercept)     22.0      3.15      6.98 0.0000647
#> 2 g2    (Intercept)     19.8      4.00      4.94 0.000801 
#> 3 g3    (Intercept)     20.6      3.79      5.43 0.000416

^{Created on 2020-04-23 by the reprex package (v0.3.0)}

Then you can use either base::write.csv() or readr::write_csv() to export the object output to a CSV file.

[amare]

Thanks so much again!.
But in my actual data, each row are repeating. You may see it in the picture.

Capture|436x199

here is the code
write.csv(IMATData %>%
pivot_longer(-c("category", "gender","Tissue","age")) %>%
group_split(name) %>%
set_names(nm = map(., ~ first(.x$name))) %>%
map(~ tidy(lm(age~ value, data = .))) %>%
map(~ filter(., p.value < 0.05)) %>%
bind_rows(.id = "var"),file = "D:/Users/amare\df.csv")
Thanks!

Actually I have to learn this amazing package.

[siddharthprabhu]

@amare I don't understand. The output of simple linear regression contains p-values for the intercept and the independent variable so you should have 2 values resulting from each model if their p-values are < 0.05. This is exactly what we see in the output.

If you're not interested in the intercept, you'll have to modify the formula for your linear model to lm(age ~ value - 1, data = .).

[amare]

Thank you so much! I get it! it was help full talking to you and other staffs in the platform!

[amare]

Hello Dear,
excuse for taking your time but I am wondering what If the categorical variable is holding more than two levels like ("A", "B"; "C"; "D").

How I should proceed If I want to see the prediction of g1, g2, and g3 on the response variable "sep" in a separate regression ("sepA vs sepC", "sepB vs sepC", "sepD vs sepC and sepD vs sepA)
I mean I want to contrast levels A, B, D with C and D with A in a separate regression.

You may use the following vectors as an example:
sep <- c("A","B", "C", "C", "D", "A","A", "D", "B","B","D";C" )
g1 <- c(35,2,3,4,5,6,7, 10,12,41,76,45)
g2 <- c(20,2,7,2,8,5,5,3,7,2,12, 13)
g3 <- c(5,0,4,5,2,4,8,9,20,1,11, 24)
data <- data.frame(sep, g1, g2, g3)

Thank you!

[nirgrahamuk]

This forum is best for tacking specific technical questions, rather than educating on broad topics. I think you should research multinomial logistic regression, there are many tutorials online for all sorts of flavours.

[amare]

Hi Siddharthprabhu,
I have tried your suggestion as follow to remove the intercept but the I found nothing in the csv file !
Capture|451x154

write.csv(IMATData %>%
pivot_longer(-c("category", "gender","Tissue","age")) %>%
group_split(name) %>%
set_names(nm = map(., ~ first(.x$name))) %>%
map(~ tidy(lm(age~ value-1, data = .))) %>%
map(~ filter(., p.value < 0.05)) %>%
bind_rows(.id = "var"),file = "D:/Users/amare\df.csv")
May I know where is the probleam?
Thanks!

[siddharthprabhu]

Maybe the model's outputs don't have any independent variables with p-values < 0.05? The code works for me with the sample data.

library(tidyverse)
library(broom)

data <- data.frame(BMI = c(30, 23, 27, 35, 12, 10, 17, 24, 21, 15, 16),
                   g1 = c(35, 2, 3, 4, 5, 6, 7, 10, 12, 41, 76), 
                   g2 = c(20, 2, 7, 2, 8, 5, 5, 3, 7, 2, 12), 
                   g3 = c(5, 0, 4, 5, 2, 4, 8, 9, 20, 1, 11))

data %>% 
  pivot_longer(-BMI) %>% 
  group_split(name) %>% 
  set_names(nm = map(., ~ first(.x$name))) %>% 
  map(~ tidy(lm(BMI ~ value - 1, data = .))) %>% 
  map(~ filter(., p.value < 0.05)) %>% 
  bind_rows(.id = "var")
#> # A tibble: 2 x 6
#>   var   term  estimate std.error statistic p.value
#>   <chr> <chr>    <dbl>     <dbl>     <dbl>   <dbl>
#> 1 g2    value     2.03     0.534      3.80 0.00347
#> 2 g3    value     1.94     0.586      3.31 0.00787

^{Created on 2020-04-24 by the reprex package (v0.3.0)}

[amare]

I am not sure what is wrong here!
If you could see the following screenshot. One is showing the result of the code with the intercept and the other is without the intercept (value-1) and it print noting.

[siddharthprabhu]

Strange. Could you please post the code snippets used to generate each of those results?

[amare]

Yes!
Here are they.

[nirgrahamuk]

if you post pictures , we cant paste those pictures into our code editors and run them.

so....

please post code as text code.
use triple backticks on a seperate line before them to format them for correct display ```

code_in_here

[amare]

Okay..sorry!

IMATData %>% 
            pivot_longer(-c("category", "gender","Tissue","age", "fatmass","weight_kg", "ffm_kg",        
                            "height_in","bmi", "GIR", "fastingGlucose", "glucose2h")) %>% 
            group_split(name) %>% 
            set_names(nm = map(., ~ first(.x$name))) %>% 
            map(~ tidy(lm(IMATData$age ~ value, data = .))) %>% 
            map(~ filter(., p.value < 0.05)) %>% 
            bind_rows(.id = "var")


IMATData %>% 
            pivot_longer(-c("category", "gender","Tissue","age", "fatmass","weight_kg", "ffm_kg",        
                            "height_in","bmi", "GIR", "fastingGlucose", "glucose2h")) %>% 
            group_split(name) %>% 
            set_names(nm = map(., ~ first(.x$name))) %>% 
            map(~ tidy(lm(IMATData$age ~ value-1, data = .))) %>% 
            map(~ filter(., p.value < 0.05)) %>% 
            bind_rows(.id = "var")

[siddharthprabhu]

Can you try replacing IMATDATA$age with age?

[nirgrahamuk]

try editing this post, and between

@amare

Okay..sorry!

@amare

IMATData %>%

put a line with three backticks. you can copy and paste these backticks if you cant find them on your keyboard. they are usually up at the top left near Tab and number 1 - ```

[amare]

I did and found the same as previous!

[amare]

where to put this line and back-ticks at the beginning or end of the code?

[siddharthprabhu]

Then I'm not really sure what's going wrong. Maybe someone else will be able to spot what I'm missing.

[nirgrahamuk]

I edited your post for you. please take a look as if you edit it yourself again to see. backticks change the mode, each time they are seen on a fresh line alone from anything else they turn the code view on orr off