use a linear model to make predictions, judge when a prediction is reasonable, and recognize data that need a nonlinear model.
Phase 4: Build
Do Now. Our rivers have gauges that measure water every few minutes. Read how they work, then think about this question.
The sleep model is y = 6x + 34. What does it predict for a student who sleeps 12 hours?
The reading opens over the lesson; close it to come back to exactly where you were.
Predict inside the data. A linear model lets us predict y for any x. Predictions inside the range of the data are called interpolation and are usually reasonable. Predictions far outside the range are called extrapolation and can be wrong or even impossible.
Worked example 1: Grace uses the sleep model. For 7.5 hours, y = 6(7.5) + 34 = 45 + 34 = 79. That is inside the 6-9.5 hour data, and the sample student who slept 7.5 hours scored 80, so 79 is reasonable. For 12 hours, the model gives 106, which is impossible on a 100-point quiz. Grace writes: “Our model is for 6 to 9.5 hours only.”
Not every pattern is a line. From Part 1, y = mx + b is linear because y changes by the same amount for each step in x. A function such as y = x2 is nonlinear: its rate of change keeps changing.
| Sample streamgage | Point 1 | Point 2 | Point 3 | Point 4 |
|---|---|---|---|---|
| Stage, or water height (feet) | 2 | 4 | 6 | 8 |
| Discharge (cubic feet per second) | 100 | 400 | 900 | 1,600 |
Worked example 2: Ben tests a line on a rating curve. The sample table above follows a curve. Ben tries a line through the first and last points, (2, 100) and (8, 1,600). Slope: 1,500 ÷ 6 = 250, and 100 = 250(2) + b gives b = −400, so y = 250x − 400. At stage 4 the line gives 600, but the table says 400. At stage 6 it gives 1,100, but the table says 900. Ben finds the rate of change for each 2-foot step: 300 ÷ 2 = 150, then 500 ÷ 2 = 250, then 700 ÷ 2 = 350 cubic feet per second per foot. The rate keeps growing, so the relationship is nonlinear, and a line misleads. The table fits y = 25x2: 25 × 22 = 100, 25 × 42 = 400, and so on.
Describe a graph in words. Ben describes the curve: “Discharge is always increasing as stage rises, and it increases faster and faster.” During a storm, a graph of stage over time might rise quickly, reach a peak, then decrease slowly as the water drains away. Words like increasing, decreasing, faster and peak tell a graph’s story.
Decide whether each prediction or model is reasonable.
Write the prediction box for your panel.
Make two predictions with your panel's model: one inside the data and one outside it. Show the work and write a one-sentence warning about the outside prediction.
Sketch a graph for this story and label the parts increasing, decreasing or constant: after school, the number of students in the library rises quickly, stays the same for an hour, then drops to zero at closing.
Use the reading. Explain why each USGS streamgage needs its own rating curve, and why one straight line could not work for every river.
Every model has a range where it works. Good analysts say where that is.
Exit ticket: how can you tell from a table whether a relationship is linear or nonlinear? Use the rating table as your example.
Pacing: warm-up and reading 7 minutes, mini lecture 10, guided practice 14, independent 7, closure 2.
Likely misconceptions: believing a model works for any input; calling a curve linear because it is always increasing; reading a graph of stage over time as a picture of a hill.
Supports: a range bar drawn under each scatter plot (the data’s smallest to largest x). Stretch: show that the rate of change in y = 25x2 grows by 100 for each 2-foot step (150, 250, 350 per foot means 300, 500, 700 per step).
Note: the rating table is sample data in a simple square pattern, chosen for grade 8; real USGS ratings are not given on the reading page. Local link: Pittsburgh’s rivers have USGS streamgages, and the CO2 model numbers come from 1.17.
Answer guide: warm-up 106 (impossible); 79; 440 ppm; 3 hours; false; y = x2 + 1; increasing fast, then decreasing slowly. Lemonade: 2(80) − 120 = 40 cups, inside the 70-90 °F data, so trust it; 2(20) − 120 = −80 cups, which is impossible, because 20 °F is far outside the data (extrapolation). Library sketch: increasing steeply, then a flat constant piece, then decreasing to 0. Reading item: the rating depends on each channel’s shape, size, slope and roughness, and the rate of change is not constant. Exit ticket: equal steps in x should give equal changes in y for a linear table; the rating table’s changes are 300, 500 and 700.
U.S. Geological Survey, Water Science School: How Streamflow is Measured · U.S. government work, public domain
The U.S. Geological Survey, or USGS, measures the water in thousands of streams and rivers. Two numbers matter most.
The first is stage, also called gage height. It is the height of the water surface, in feet, above a set point where the stage is zero. Streamgages usually record the stage every 15 minutes.
The second is discharge, also called streamflow. It is the amount of water moving past a point in a given time, often in cubic feet per second. The page gives the rule discharge = area × velocity. Discharge is hard to measure all the time, so USGS crews measure it by hand every 6 to 8 weeks at most gages, using current meters or sound-based instruments.
Then they connect the two. For each streamgage, USGS builds a rating curve, a graph of the relation between stage and streamflow. With it, every stage reading can be turned into a discharge. The page says the rating depends on the shape, size, slope and roughness of the channel, so it is different for every streamgage. Channels change over time, from erosion, streambed material piling up, plant growth, debris and ice, so crews keep checking and updating the ratings.
Source: U.S. Geological Survey, Water Science School: How Streamflow is Measured (usgs.gov)