Looping ggplot2 multiple linear regression

Question

I am trying to loop my multiple linear regression plot and summaries, but I keep encountering an error in R that states Error: More than one expression parsed. I am not sure how to fix this or if there is a better way to achieve what I want to do which is mainly:

1. Plot a multiple linear regression plot with Group as the colour
2. Get summary for each of the linear regression lines based on Group
3. Compute regression summary
4. Perform anova to determine differences

colNames <- names(df)[c(35:39)]
for(i in colNames){
  plt <- ggplot(df, 
aes_string(x=df$MachineLength, y=i, fill=df$Group, color=be_nlyl$Group)) + 
geom_smooth(method=lm) + 
geom_point(size = 2, alpha=0.7) + 
labs(title="Machine", subtitle = "Machine Type") + 
theme_bw() + 
theme(plot.title = element_text(hjust=0.5, face="bold"), 
plot.subtitle = element_text(hjust=0.5))
  print(plt)
  lm_A <- lm(formula = i ~ MachineLength, data = subset(be_nlyl, Group == "A"))
  summary(lm_A) %>% print()
lm_B <- lm(formula = i ~ MachineLength, data = subset(be_nlyl, Group == "B"))
  summary(lm_B) %>% print()
  clz.lm <- lm(formula = i ~ Group + MachineLength + Group:MachineLength, data = df)
summary(clz.lm) %>% print()
  ano.lm <- Anova(lm(i ~ MachineLength*Group, data = df))
  print(ano.lm)
}

Anyone have ideas of how to implement above? Thank you!

Answer 1

Try the following :

1. Create lists of length colNames to store all the outputs so that instead of just printing the output we can store them as well.

2. Use for loop over the index of colNames instead of actual column names so that you can use that as an index to store the output for different objects.

3. aes_string has be deprecated so we use .data pronoun to pass column name as variable.

4. Use sprintf to create formula string which is passed in lm function.

library(ggplot2)

colNames <- names(df)[c(35:39)]
plt <- vector('list', length(colNames))
lm_A <- vector('list', length(colNames))
summary_lm_A <- vector('list', length(colNames))
summary_lm_B <- vector('list', length(colNames))
lm_B <- vector('list', length(colNames))
clz.lm <- vector('list', length(colNames))
summary_clz.lm <- vector('list', length(colNames))
ano.lm <- vector('list', length(colNames))

for(i in seq_along(colNames)) {
  var <- colNames[i]
  plt[[i]] <- ggplot(df, aes(MachineLength, .data[[var]], fill= Group, color= Group)) + 
               geom_smooth(method=lm) + 
               geom_point(size = 2, alpha=0.7) + 
               labs(title="Machine", subtitle = "Machine Type") + 
               theme_bw() + 
               theme(plot.title = element_text(hjust=0.5, face="bold"), 
                     plot.subtitle = element_text(hjust=0.5))
  lm_A[[i]] <- lm(sprintf('%s~MachineLength', var), data = subset(df, Group == "A"))
  summary_lm_A[[i]] <- summary(lm_A[[i]])
  lm_B[[i]] <- lm(sprintf('%s~MachineLength', var), data = subset(df, Group == "B"))
  summary_lm_B[[i]] <- summary(lm_B[[i]])
  clz.lm[[i]] <- lm(sprintf('%s~Group + MachineLength + Group:MachineLength', var), data = df)
  summary_clz.lm[[i]] <- summary(clz.lm[[i]])
  ano.lm[[i]] <- Anova(lm(sprintf('%s~MachineLength*Group', var), data = df))
}