Five steps: enter → choose → run → read → use: At SL the only regression you run is the LINEAR one: the regression line of y on x (topic 4.4).
A curved model at SL gets its numbers by solving equations instead (see "Finding the parameters").
GDC regression workflow (TI-84)
- STAT → EDIT → enter x-values in L1, y-values in L2.
- STAT → CALC → 4:LinReg(ax+b).
- Confirm lists L1 and L2, then press ENTER.
- Read a, b and r.
- Store to Y1 via "RegEQ" (or type manually) to evaluate and graph.
Example — run regression and predict [5 marks]
A shop records the price p ($) it charges for a phone case and the number of cases q sold per week for five different prices:
p: 6, 8, 10, 12, 14
q: 92, 79, 66, 53, 40
(a) Use your GDC to find the equation of the regression line of q on p, giving each coefficient to 3 significant figures.
(b) Use your equation to estimate the number of cases sold per week when the price is $11.
Step by step
- (a) Enter the prices in L1 and the sales in L2, then run LinReg(ax+b) on the GDC. The calculator returns the gradient a and intercept b.
- Write the line in the form q = ap + b, rounding each coefficient to 3 s.f.
- (b) Substitute p = 11 into the regression equation.
- Sales are whole cases, and p = 11 lies inside the data (interpolation), so round sensibly.
Final answer
(a) q = −6.50p + 131. (b) About 60 cases per week.
GDC walkthrough
Step through the exact calculator keystrokes, screen by screen, in study mode.
Free preview
This is the free notes preview
You're reading the free notes. Aimnova Pro unlocks the full study experience — and you can try it with your first topic free to keep:
- FlashcardsLock in vocabulary and key terms with spaced repetition.
- Practice questionsAnswer exam-style questions and get instant AI marking.
- Mock exams & past-paper vaultSit full mocks and see exactly how examiners award marks.
- Personalised study planA daily plan built around your exam date and weak areas.
Look at the scatter plot shape first: Before running any regression, plot the data (STAT PLOT on TI; StatGraph on Casio) and look at the shape.
The shape tells you which model to try.
| Scatter plot shape | Model to try | Key signal in question |
|---|---|---|
| Straight line | Linear: y = mx + c | "constant rate", "per unit", r close to ±1 |
| Single peak or valley | Quadratic: y = ax² + bx + c | "maximum", "minimum", projectile |
| Rapid increase, levels off | Exponential: y = k × aˣ | "percentage growth/decay", "doubles every..." |
| Curve through origin, increasing | Power: y = axⁿ | "proportional to square/cube", "directly proportional to" |
| Repeating up-down pattern | Sinusoidal: y = a sin(bx) + d | "tide", "temperature cycle", "Ferris wheel" |
The question often names the model: Questions usually say "is modelled by h(x) = ax² + bx + c" or "N = k × aᵗ".
Then you only need its parameters. At SL they come from solving equations (one per point), never from a curved regression. Only a straight line is found by regression.
Memorize terms 3x faster
Smart flashcards show you cards right before you forget them. Perfect for definitions and key concepts.
Copy coefficients exactly — then write the equation: After running regression, the GDC displays coefficients.
Write the full equation immediately — do not rely on memory.
Round coefficients to 3 significant figures unless the question specifies otherwise.
Xavie collects apartment prices y (millions $) and distances x (km) from city centre.
GDC LinReg gives: a = −0.0693, b = 3.10, r = −0.998.
Write the regression equation and use it to predict y when x = 15.
Step by step
- Write the linear regression equation.
- Comment on r: r = −0.998 is very close to −1 → strong negative linear correlation. The linear model is appropriate.
- Predict y when x = 15 (within data range, so interpolation).
Final answer
y = −0.0693x + 3.10. Predicted price at 15 km from centre ≈ $2.06 million.
r says how straight: Pearson's r (on the same LinReg screen) says how close the points lie to a straight line. It runs from −1 to +1 and only judges a LINEAR model.
| r | Strength | What it means |
|---|---|---|
| +1 or −1 | Perfect | Every point lies on one straight line. |
| close to +1 or −1 (e.g. 0.95, −0.96) | Strong | A line is a good model. |
| close to 0 (e.g. 0.12) | Weak or none | A line is a poor model. |
What to write when commenting on r: IB mark scheme expects: (1) state the value of r, (2) describe the strength (strong / moderate / weak), (3) state the direction (positive / negative).
Example: "r = −0.998 indicates a strong negative linear correlation between distance and apartment price."
Example — comment on the correlation coefficient [3 marks]
For two sets of bivariate data, a GDC reports the following correlation coefficients for a linear regression.
Data set A: r = −0.96
Data set B: r = 0.38
(a) Describe the correlation in data set A.
(b) State, with a reason, whether a linear model is appropriate for data set B.
Step by step
- (a) When you comment on r, give three things: the strength, the direction, and that it is linear. Here r = −0.96 is very close to −1.
- So data set A shows a strong negative linear correlation — as one variable increases, the other tends to decrease, and the points lie close to a straight line.
- (b) For data set B, r = 0.38 is close to 0, which means the points are widely scattered about any straight line. A linear model is therefore a poor fit and not appropriate.
Final answer
(a) Strong negative linear correlation. (b) No — r = 0.38 is close to 0, so the data is weakly correlated and a linear model is not appropriate.
Know your predicted grade
Take timed mock exams and get detailed feedback on every answer. See exactly where you're losing marks.
One point, one equation: Put each point you know, (x, y), into the model. Each point gives one equation.
Three unknown letters need three equations; solve them together on the GDC. At SL you never run a curved regression: this is how a curved model gets its numbers.
Example: a quadratic through three points
The path of a ball is modelled by h(x) = ax² + bx + c, where h metres is its height x metres along the ground. The ball passes through (0, 1.5), (10, 6.5) and (20, 7.5). Find a, b and c.
Step by step
- Put in (0, 1.5): everything with x vanishes.
- Put in (10, 6.5).
- Put in (20, 7.5).
- Solve the three equations together on the GDC.
Final answer
h(x) = −0.02x² + 0.7x + 1.5
See it drawn, step by step
In My Learning, handwritten diagrams build one step at a time. Play, pause and replay each explanation until it clicks. Create your free account to try it.
GDC walkthrough
Step through the exact calculator keystrokes, screen by screen, in study mode.
Example: an exponential model from a starting value
The number of members of a running club t years after it opened is modelled by N = k × aᵗ. It opened with 500 members and had 820 members after 4 years. Find k and a.
Step by step
- At the start t = 0, and a⁰ = 1, so N = k.
- Use the second value: t = 4, N = 820.
- Divide by 500, then take the fourth root.
Final answer
k = 500, a = 1.13: N = 500 × 1.13ᵗ
Shortcut next time: The value at the start (t = 0 or x = 0) gives the constant straight away: c for a quadratic, k for an exponential. Every other point then gives one more equation.
Let the data choose the model: Look at equal steps in x (or time):
• equal DIFFERENCES → linear
• equal RATIOS (the same factor each step) → exponential
• a rise then a fall → quadratic
Then check the model makes sense in the context.
Example: linear or exponential?
A cup of tea cools in a room at 20 °C. Its temperature T °C, every 5 minutes, is 90, 70, 56, 46, 39. (a) State, with a reason, whether a linear or an exponential model fits better. (b) Explain why a linear model is not reasonable for large times.
Step by step
- (a) The drops are 20, 14, 10, 7: they get smaller, so the data are not linear.
- Above room temperature, T − 20 is 70, 50, 36, 26, 19: it falls by about the same factor each time.
- (b) A straight line keeps falling forever.
Final answer
(a) Exponential: the decreases get smaller (a constant ratio). (b) A line would predict temperatures below 20 °C; the tea cannot cool below the room.
A reasonable domain: Keep x where the model makes sense. The ball h(x) = −0.02x² + 0.7x + 1.5 is in the air only until h = 0: the GDC gives x = 37.0 m, so a reasonable domain is 0 ≤ x ≤ 37.0.
Predicting inside the data is safer (interpolation). Far outside it (extrapolation), the model may stop working.