Contents
Statistics and data

Linear regression: two columns, then a, b and r

Pair x with y row by row, choose y = a + bx, and read the intercept, the slope and the correlation off the same table.

8 min read Intermediate updated 2026-11-07

Regression on the fx-991EX is the Statistics table with a second column. You enter x and y as pairs, choose y = a + bx, and the machine hands back the intercept, the slope, r and r². The pairing is positional: the third x belongs to the third y, so one skipped row quietly changes the model without any warning on screen.

Enter the pairs

The example below is a small data set with clean numbers, so every result can be checked by hand afterwards: x = 1, 2, 3, 4, 5, 6 with y = 3, 5, 4, 7, 8, 9.

Keystrokes

  1. 1
    MENU

    Open the application menu.

  2. 2
    ▼ ▶

    Move the highlight onto the Statistics icon.

  3. 3
    =

    Open the application.

  4. 4
    ▼ =

    Choose the regression entry labelled 2-Variable, or y = a + bx. Follow the label rather than a number: the item order differs between EX models.

  5. 5
    1 = 2 = 3 = 4 = 5 = 6 =

    The x column, top to bottom. Each = confirms the value and moves down one row.

  6. 6
    ▶

    Move to the y column. Do not type the y values before this.

  7. 7
    3 = 5 = 4 = 7 = 8 = 9 =

    The matching y values in the same row order: 3, 5, 4, 7, 8, 9.

  8. 8
    AC

    Leave the table. Both columns stay inside the application.

fx-991EX Plus 0.
Open Statistics with MENU → ▼ → =, cross to the y column with ▶, leave with AC. Rendered from the same layout data the online calculator uses — key positions and silkscreen labels are not stylised.

Read a, b and r

Open the Statistics application again and choose the calculated results rather than the data table. Press = or ▼ to walk down the list.

ItemValueWhat it is
A1.8the intercept: the fitted y when x = 0
B1.2the slope: y rises by 1.2 for every +1 in x
x̄ / ȳ3.5 / 6the two column means
r0.9486832981correlation coefficient, always between −1 and 1
r²0.9the share of the spread in y that the line explains

Check the line by hand

The results list also gives the sums, which is enough to rebuild both coefficients: Σx = 21, Σy = 36, Σx² = 91 and Σxy = 147. Then b = (nΣxy − ΣxΣy) ÷ (nΣx² − (Σx)²) = (6 × 147 − 21 × 36) ÷ (6 × 91 − 21²) = 126 ÷ 105 = 1.2, and a = ȳ − b x̄ = 6 − 1.2 × 3.5 = 1.8.

  • The fitted line always passes through (x̄, ȳ). Put 3.5 in for x: 1.8 + 1.2 × 3.5 = 6, exactly the mean of the y column. A hand-computed line that misses that point is arithmetic, not rounding.
  • r is signed, r² is not: r = 0.9486832981 here, and its square is the 0.9 on the screen. Quote r when the question asks for correlation and r² when it asks how well the model fits.
  • Predict x = 7: 1.8 + 1.2 × 7 = 10.2. That is one step outside the entered range, so most mark schemes want a note saying it is an extrapolation.
Example The fitted value at x = 7

Input

y = 1.8 + 1.2 × 7

Result

10.2

Key the equation with the digits the screen printed. Rounding 1.2 or 1.8 first moves the prediction.

Was this guide helpful?

Tell us what was unclear, missing or wrong. A concrete sentence is more useful than a rating.