Regression and correlation worksheet with answer key

8 problems on regression and correlation, ready to print for class or homework. The answer key prints on its own page, so hand out everything before it.

Problem 1

A greenhouse trial links the weekly fertilizer dose (in grams) given to each tomato plant, xx, to the season's yield (in kilograms), yy. Across the plants the summary statistics are xˉ=10\bar{x} = 10 grams, yˉ=50\bar{y} = 50 kg, sx=2s_x = 2 grams, sy=8s_y = 8 kg, and correlation r=0.75r = 0.75. Find the least-squares regression line y^=a+bx\hat{y} = a + bx and interpret the slope in context.

Problem 2

A university makerspace logs, for four one-hour blocks, the number of 3D printers running, xx, and the electricity used that hour in kilowatt-hours, yy: (2,5)(2, 5), (4,9)(4, 9), (6,11)(6, 11), (8,15)(8, 15). Find the least-squares regression line y^=a+bx\hat{y} = a + bx from the data.

Problem 3

A candle maker fit a least-squares line predicting a candle's burn time (in hours) from its weight (in ounces): y^=2+1.5x\hat{y} = 2 + 1.5x, where xx is the weight and y^\hat{y} is the predicted burn time. (a) An 8-ounce candle actually burns for 15 hours. Find its residual and say what the sign means. (b) A 10-ounce candle has a residual of −2-2 hours. Find its observed burn time.

Problem 4

A cycling shop fit a least-squares line predicting a used e-bike's resale price (in dollars) from its odometer reading (in hundreds of kilometers), xx: y^=1500−50x\hat{y} = 1500 - 50x. The e-bikes in the data had odometer readings from x=5x = 5 to x=40x = 40 (that is, 500 to 4000 km), and the correlation was r=−0.9r = -0.9. Interpret the slope, the intercept, the correlation rr, and the coefficient of determination r2r^2 in context.

Problem 5

A call center studies its shifts. Across many shifts the number of agents on duty, xx, has mean xˉ=20\bar{x} = 20 and standard deviation sx=5s_x = 5, while the calls resolved per hour, yy, has mean yˉ=60\bar{y} = 60 and standard deviation sy=15s_y = 15; the correlation is r=0.8r = 0.8. (a) Find the least-squares regression line. (b) One shift had 25 agents and resolved 80 calls in an hour. Find that shift's residual and interpret it.

Problem 6

An urban planner collects data on many neighborhoods and finds a correlation of r=0.7r = 0.7 between the number of coffee shops in a neighborhood, xx, and the average monthly apartment rent, yy. The data come from an observational study. (a) Interpret rr in context. (b) Compute and interpret the coefficient of determination r2r^2. (c) Does this show that adding coffee shops causes rents to rise? Explain.

Problem 7

A marine biologist measures sea turtles and summarizes shell length (in centimeters), xx, and body weight (in kilograms), yy: xˉ=40\bar{x} = 40 cm, sx=10s_x = 10 cm, yˉ=30\bar{y} = 30 kg, sy=20s_y = 20 kg, and r=0.8r = 0.8. The measured shells ranged from 30 to 55 cm. (a) Find the least-squares regression line. (b) Interpret the slope. (c) Interpret the intercept, and explain any problem with it. (d) Find r2r^2 and interpret it. (e) Predict the weight of a turtle with a 50 cm shell, and state whether that prediction is interpolation or extrapolation.

Problem 8

An ecologist runs a linear regression predicting a lake's dissolved oxygen (in mg/L), yy, from the water temperature (in degrees Celsius), xx. The output gives the line y^=14−0.25x\hat{y} = 14 - 0.25x with correlation r=−0.95r = -0.95. (a) Interpret the slope. (b) At 20 degrees Celsius the measured dissolved oxygen was 8.5 mg/L; find and interpret the residual. (c) Find r2r^2 and interpret it. (d) The residual plot shows a clear U-shaped pattern. Given the high r2r^2, what does that say about using a line here?

Regression and correlation: answer key

  1. The least-squares line is y^=20+3x\hat{y} = 20 + 3x; each additional gram of weekly fertilizer predicts a 3 kg increase in yield.

  2. The least-squares line is y^=2+1.6x\hat{y} = 2 + 1.6x, so each extra printer running predicts 1.6 more kilowatt-hours that hour.

  3. (a) The residual is +1+1 hour, so the model underpredicted the burn time by 1 hour. (b) The observed burn time is 17+(−2)=1517 + (-2) = 15 hours, and the negative residual means the model overpredicted.

  4. Slope: each extra 100 km predicts a 50 dollar drop. Intercept: 1500 dollars predicted at 0 km, an extrapolation. r=−0.9r = -0.9 is a strong negative association, and r2=0.81r^2 = 0.81 means about 81 percent of the price variation is explained by odometer reading.

  5. (a) y^=12+2.4x\hat{y} = 12 + 2.4x. (b) The residual is +8+8 calls; with 25 agents the line predicted 72 calls, and the shift resolved 8 more than predicted.

  6. (a) r=0.7r = 0.7 is a moderately strong positive linear association. (b) r2=0.49r^2 = 0.49, so about 49 percent of the rent variation is explained by coffee-shop counts. (c) No; the study is observational, and a lurking variable such as neighborhood desirability could drive both.

  7. (a) y^=−34+1.6x\hat{y} = -34 + 1.6x. (b) Each extra cm of shell predicts 1.6 kg more weight. (c) The intercept −34-34 kg is a negative weight and an extrapolation, so it has no real meaning. (d) r2=0.64r^2 = 0.64, so 64 percent of the weight variation is explained. (e) y^=46\hat{y} = 46 kg, an interpolation since 50 cm is within 30 to 55 cm.

  8. (a) Each extra degree predicts a 0.25 mg/L drop. (b) Predicted 9 mg/L, so the residual is 8.5−9=−0.58.5 - 9 = -0.5 mg/L and the model overpredicted by 0.5. (c) r2=0.9025r^2 = 0.9025, about 90.3 percent of the variation explained. (d) The U-shaped residual plot shows curvature, so a line is not appropriate despite the high r2r^2.

The key gives final answers only. The full worked solution for each problem is on the matching practice set, and every other sheet is on the worksheets page.

More worksheets from this unit