Linear Regression and log-log Transformation, R

Linear Regression and log-log Transformation

#R version 3.3.2 
# Let's see the variables

head(cars) 
#Let's see descriptive statistics 
summary(cars)
#Let's see standard deviation as well
print(apply(cars,2,sd))
#do te same thing for log transformation of variables
lncars <- log(cars)
summary(lncars)
print(apply(lncars,2,sd))
#plot variables 
scatter.smooth(x=cars$speed, y=cars$dist, main="Dist ~ Speed")

#run a regressio using original data 
original <- lm(dist ~ speed, data=cars)
summary(original)
#get predicted y 
cars$predicted <- predict(original)
#Plot actual vs predicted

library(ggplot2)
ggplot(cars, aes(x = speed, y = dist)) +
geom_smooth(method = "lm", se = FALSE, color = "lightgrey") +  # Plot regression slope
  geom_segment(aes(xend = speed, yend = predicted), alpha = .2) +  # alpha to fade lines
  geom_point() +
  geom_point(aes(y = predicted), shape = 1) +
  theme_bw() 
#run a regressio using log transformation of  data 
scatter.smooth(x=lncars$speed, y=lncars$dist, main="lnDist ~ lnSpeed")
loglog <- lm(dist ~ speed, data=lncars)
summary(loglog)
#get predicted y 
lncars$predicted <- predict(loglog)

#Plot actual vs predicted 
ggplot(lncars, aes(x = speed, y = dist)) +
  geom_smooth(method = "lm", se = FALSE, color = "lightgrey") +  # Plot regression slope
  geom_segment(aes(xend = speed, yend = predicted), alpha = .2) +  # alpha to fade lines
  geom_point() +
  geom_point(aes(y = predicted), shape = 1) +
  theme_bw()


run \| edit \| history \| help	0

λ

                                      .NET NoSQL database for rapid development