1

I am trying to plot multiple variable correlation plots on one page - but the points aren't being plotted correctly when I use a for loop. They are fine if I create each plot individually.

I have a dataframe called sig_df with the variable names, intercept and slope of the regression line, and the p-value of the correlation: 

> sig_df
                 x                 y  intercept     slope      p_value
1 Share.incentives     User.profiles -0.1667891 0.5826598 0.0003577163
2 Share.incentives       Translation  2.2689199 0.4790595 0.0005928465
3      Translation  User.interaction  1.4696970 0.5757576 0.0008921255

and also a master dataframe with the x and y values of each variable, for a sample of 26

> master
   Share.incentives User.profiles Translation User.interaction
1                 5             4           5                5
2                 4             1           3                4
3                 6             4           5                5
4                 6             4           4                4
5                 4             1           5                3
..              ...           ...         ...              ...

I've used a for loop to look at every row in sig_df, look up the variable names in master and make a correlation plot. I have stored these plots in a list, then use grid.arrange to plot them on a page - the ablines are plotting correctly but the points on every plot are the same as the final plot in my list! When I don't use a for loop, this doesn't happen. Does anyone know what could be causing this behaviour? 

plot_list<- list()

for(i in 1:nrow(sig_df)){
  y <- (sig_df$y[[i]])
  x <- (sig_df$x[[i]])
  slope <- (sig_df$slope[[i]])
  intercept <- (sig_df$intercept[i])
  plot_list[[i]] <- ggplot(data=master, aes(get(x), get(y))) + geom_point() + geom_abline(slope=slope, intercept=intercept) 

}

do.call(grid.arrange, c(plot_list, ncol=4))

Thanks, KP

K-P
  • 79
  • 2
  • 5
  • 1
    I've now overcome this issue by selecting the chosen columns in sig_df by avoiding the use of 'get': `df <- select(master, get(x), get(y)) colnames(df)<- c('x','y') plot_list[[i]] <- ggplot(data=df, aes(x,y)) + geom_point() + geom_abline(slope=slope, intercept=intercept) ` – K-P Sep 08 '15 at 15:30
  • that's the correct approach; `get()` is generally best avoided, and definitely not recommended within a ggplot. Note that you can simplify the grid.arrange step with `grid.arrange(grobs = plot_list, ncol=4)` – baptiste Sep 08 '15 at 19:44

1 Answers1

1

I encountered similar problem. Here's my solution: change the mapping parameter to aes_string() in funtion ggplot

  plot_list[[i]] <- ggplot(data=master, aes_string(get(x), get(y))) + geom_point() + geom_abline(slope=slope, intercept=intercept)

Hope this helps.

jiang ziqi
  • 47
  • 1
  • 1
  • 5