Lab
Model Builder
Pick a data set and a model family. The lab fits the model by least squares, the way a graphing calculator's regression does, and plots the residuals underneath. Before you switch families, look at the scatterplot and say which one you expect to fit, then check whether the residual plot has a pattern. You can type over any value in the table, add rows or remove them. The data sets are invented.
Challenges
- Load the candle data and fit a linear model. How fast does the model say the candle gets shorter, and when does it predict the candle burns out?
- Load the ball toss data and fit a quadratic model. When does the model say the ball is highest, how high is it then, and when does the model say it lands?
- Load the used car data and fit an exponential model. By what percent does the model say the car loses value each year, and what does it predict the car is worth at 12 years old?
- Load the town population data. Fit a linear model and look at the residual plot, then fit an exponential model and look again. Which residual plot shows a pattern, and which model would you choose?
- Keep the town data and go back to the linear model. For which years does the model underestimate the population?
- Load the typing practice data and fit a logarithmic model. What does it predict after 20 weeks, and after how many weeks does it predict 65 words per minute?