Overview
Cubic regression fits an S-shaped curve to your data, for patterns that bend more than once.
Paste your x-y points, one per line. Add an x value if you want a prediction.
Worked examples
| Inputs | Result |
|---|---|
| Your x-y points1, 2.1 · 2, 3.9 · 3, 9.2 · 4, 15.8 · 5, 25.3 · 6, 35.9 | y = −0.066667x³ + 1.7857x² − 2.8619x + 3.2R²: 0.9998 from 6 points |
How it works
The curve is chosen by least squares, which makes the squared vertical distances from the points to the curve as small as possible. R² shows the share of the variation in y the curve explains.
The formula
| a, b, c, d | Coefficients chosen to fit your points |
Practical tips
- You need at least four points with different x values.
- Cubic curves swing sharply outside your data, so don't trust predictions far beyond it.
- Adding terms always raises R², so a higher score doesn't always mean a better model.
Frequently asked questions
When should I use cubic regression?
When the data clearly bend in two directions and a parabola misses the pattern.
Can a cubic fit be misleading?
Yes. With few points it can follow noise rather than the real trend.
How many points do I need?
At least four, and ideally many more.