# Simple non-linear regression problem

I'm trying to model a simple use case: predicting the price of a car based on its mileage, with RStudio. I know it's a really naive model, just one variable, but it's for comprehension purposes.

My first attempt was to to use the lm function:

predictions <- lm(price~mileage, data = ads_clean)


If I plot the model using the visreg function, I get a scatter plot of my prices/mileages with a straight line (negative slope) on it. I can see according to that plot that I can obtain negative predictions (it seems normal according to the negative coefficient of the mileage).

The second attempt was to elminate such negative predictions using a log10 on the price. What I'm predicting now is not the price, but the log10(price). If I want to get back to the 'right' predicted price I use 10^(predictedPrice).

predictions <- lm(log10(price)~mileage, data = ads_clean)


If I plot the model I still get a straight line on my scatter plot, but without negative predictions this time.

How do I get a curve instead of a straight line? I suppose that lm can only generate straight lines (ax1 + bx2 + .... + A).

May I use another kind of function? glm?

I'd like to get such visreg (red curve):

• Could you provide a data sample as an example? – Tim Mar 19 '15 at 10:34
• These data look overcool! Is this public data? – Elvis Mar 19 '15 at 11:33
• Re "How to get a curve instead of a straight line?" All the lines you have drawn are curves. The blue line fit to the log-log data will, when drawn in the original units, will even look curved. – whuber Mar 19 '15 at 19:12

If you log-transformed your outcome variable and then fit a regression model, just exponentiate the predictions to plot it on the original scale.

In many cases, it's better to use some nonlinear functions such as polynomials or splines on the originale scale, as @hejseb mentioned. This post might be of interest.

Here is an example in R using the mtcars dataset. The variable used here were chosen totally arbitrarily, just for illustration purposes.

First, we plot Log(Miles/Gallon) vs. Displacement. This looks approximately linear.

After fitting a linear regression model with the log-transformed Miles/Gallon, the prediction intervals on the log-scale look like this:

Exponentiating the prediction intervals, we finally get this graphic on the original scale:

This ensures that the prediction intervals never go below 0.

We could also fit a quadratic model on the original scale and plot the prediction intervals.

Using a quadratic fit on the original scale, we cannot be sure that the fit and prediction intervals stay above 0.

Here is the R-code that I used to generate the figures.

#------------------------------------------------------------------------------------------------------------------------------
#------------------------------------------------------------------------------------------------------------------------------

data(mtcars)

#------------------------------------------------------------------------------------------------------------------------------
# Scatterplot with log-transformation
#------------------------------------------------------------------------------------------------------------------------------

plot(log(mpg)~disp, data = mtcars, las = 1, pch = 16, xlab = "Displacement", ylab = "Log(Miles/Gallon)")

#------------------------------------------------------------------------------------------------------------------------------
# Linear regression with log-transformation
#------------------------------------------------------------------------------------------------------------------------------

log.mod <- lm(log(mpg)~disp, data = mtcars)

#------------------------------------------------------------------------------------------------------------------------------
# Prediction intervals
#------------------------------------------------------------------------------------------------------------------------------

newframe <- data.frame(disp = seq(min(mtcars$disp), max(mtcars$disp), length = 1000))

pred <- predict(log.mod, newdata = newframe, interval = "prediction")

#------------------------------------------------------------------------------------------------------------------------------
# Plot prediction intervals on log scale
#------------------------------------------------------------------------------------------------------------------------------

plot(log(mpg)~disp
, data = mtcars
, ylim = c(2, 4)
, las = 1
, pch = 16
, main = "Log scale"
, xlab = "Displacement", ylab = "Log(Miles/Gallon)")

lines(pred[,"fit"]~newframe$disp, col = "steelblue", lwd = 2) lines(pred[,"lwr"]~newframe$disp, lty = 2)
lines(pred[,"upr"]~newframe$disp, lty = 2) #------------------------------------------------------------------------------------------------------------------------------ # Plot prediction intervals on original scale #------------------------------------------------------------------------------------------------------------------------------ plot(mpg~disp , data = mtcars , ylim = c(8, 38) , las = 1 , pch = 16 , main = "Original scale" , xlab = "Displacement", ylab = "Miles/Gallon") lines(exp(pred[,"fit"])~newframe$disp, col = "steelblue", lwd = 2)
lines(exp(pred[,"lwr"])~newframe$disp, lty = 2) lines(exp(pred[,"upr"])~newframe$disp, lty = 2)

#------------------------------------------------------------------------------------------------------------------------------
# Quadratic regression on original scale
#------------------------------------------------------------------------------------------------------------------------------

quad.lm <- lm(mpg~poly(disp, 2), data = mtcars)

#------------------------------------------------------------------------------------------------------------------------------
# Prediction intervals
#------------------------------------------------------------------------------------------------------------------------------

newframe <- data.frame(disp = seq(min(mtcars$disp), max(mtcars$disp), length = 1000))

pred <- predict(quad.lm, newdata = newframe, interval = "prediction")

#------------------------------------------------------------------------------------------------------------------------------
# Plot prediction intervals on log scale
#------------------------------------------------------------------------------------------------------------------------------

plot(mpg~disp
, data = mtcars
, ylim = c(7, 36)
, las = 1
, pch = 16
, main = "Original scale"
, xlab = "Displacement", ylab = "Miles/Gallon")

lines(pred[,"fit"]~newframe$disp, col = "steelblue", lwd = 2) lines(pred[,"lwr"]~newframe$disp, lty = 2)
lines(pred[,"upr"]~newframe\$disp, lty = 2)


If all you want is a quadratic term, you can use lm(y~x+I(x^2)). An example:

For your model that would mean predictions <- lm(price~mileage+I(mileage^2), data = ads_clean). For higher order polynomials, you can just add them in the same way. You could also try some nonparametric regression, for example locpoly.

x <- rnorm(100)
y <- x + x^2 + rnorm(100)
plot(x, y)
model1 <- lm(y~ x+ I(x^2))
plotdata <- cbind(x, predict(model1))
lines(plotdata[order(x),], col = "red")


Please be aware that, depending on your goal, this might be associated with other problems such as heteroscedasticity. If you want to make inference, you need to pay extra care that the assumptions you would rely on actually appear to be satisfied. But, if you are truly only interested in how to get a curve instead of a straight line and you're just playing around, this is sufficient.

• All of your recommendations will suffer from heteroscedasticity with the example dataset in the question. – Roland Mar 19 '15 at 11:54
• @Roland, sure, but the question was how to "get a curve instead of a straight line". Judging by the first line of the post, finding a good model or make correct inference does not seem to be the objective. – hejseb Mar 19 '15 at 11:58
• On this site, even if OP asks for it, you should not give advice that would get them into trouble; at least not without a warning. – Roland Mar 19 '15 at 12:01
• @Roland if that were a formal guideline I'd have been banned by now – shadowtalker Mar 19 '15 at 12:25
• @Roland, can you provide a link to resources to that would assist the OP (and others) to understand and deal with heteroscedasticity? – Tony Ladson Mar 24 '15 at 22:27