MPG Predictor for mtcars Data

Using Weight, Horsepower and Transmission Type

Lou Solot

Method

  1. Use Linear Regression to find the variables that can predict MPG
  2. Build a prediction model using those variables
  3. Create a shiny app that enables users to input those variables in order to predict MPG
  4. Include the mtcars table so that users can compare their result to the real data
sort(cor(mtcars)[1,])

##         wt        cyl       disp         hp       carb       qsec 
## -0.8676594 -0.8521620 -0.8475514 -0.7761684 -0.5509251  0.4186840 
##       gear         am         vs       drat        mpg 
##  0.4802848  0.5998324  0.6640389  0.6811719  1.0000000

This correlation shows that wt, hp, am, cyl and disp are all highly correlated with mpg

Eliminate Variables Highly correlated with Each Other

sort(cor(mtcars)[4,])

##        mpg         vs       qsec       drat         am       gear 
## -0.7761684 -0.7230967 -0.7082234 -0.4487591 -0.2432043 -0.1257043 
##         wt       carb       disp        cyl         hp 
##  0.6587479  0.7498125  0.7909486  0.8324475  1.0000000

We can eliminate hp and disp since predictors should not be highly correlated with each other

Build the Model Using Weight, Horsepower and Transmission Type

pred <- lm(mpg ~ am + wt + hp, data = mtcars)
summary(pred)$coef

##                Estimate  Std. Error   t value     Pr(>|t|)
## (Intercept) 34.00287512 2.642659337 12.866916 2.824030e-13
## am           2.08371013 1.376420152  1.513862 1.412682e-01
## wt          -2.87857541 0.904970538 -3.180850 3.574031e-03
## hp          -0.03747873 0.009605422 -3.901830 5.464023e-04

Plots

par(mfrow = c(2,2))
plot(pred)

plot of chunk unnamed-chunk-4

Conclusion

summary(pred)$r.squared

## [1] 0.8398903
  • This model explains ~84% of the variance
  • This shows that the combination of weight, horsepower and transmission type is a good predictor of mpg