Exponential Regression Calculator

What the exponential regression calculator does

Exponential regression fits a curve of the form y equals a times b to the power x through your data. This calculator takes your x and y points, works out the two coefficients, gives the model equation, and plots the curve against your points. It is the tool for data that grows or shrinks by a constant factor.

Where a straight line adds the same amount each step, an exponential multiplies by the same factor each step. Below is when that fits and how the curve is found.

How to use it

1. Enter your data as x,y pairs in the box, separated by spaces or new lines, at least three of them.
2. Press Calculate for the coefficients, the equation, and a plot, or Reset to clear it.

The y values need to be positive, since the method works through their logarithms.

When an exponential fits better than a line

The clue is in how the data changes. If y goes up by a roughly constant amount for each step in x, a straight line fits. If y instead multiplies by a roughly constant factor each step, doubling, or growing by ten percent, or halving, then the relationship is exponential and a line will fit poorly.

This is the pattern of compound growth and decay: populations, savings with compound interest, the spread of something through a network, or radioactive material dwindling away. On these, a line undershoots then overshoots, because it cannot bend the way the data does. The exponential curve bends to match, capturing growth that feeds on itself.

How the curve is fitted

There is a neat trick behind it. Fitting a curve directly is awkward, so the calculator takes the logarithm of each y value first. That step turns an exponential relationship into a straight-line one, because the logarithm of a times b to the x is a straight line in x.

So the calculator fits an ordinary least squares line to x against the logarithm of y, then converts the line's slope and intercept back into the original coefficients. This log-transform approach is the standard way to fit an exponential, turning a curved problem into the straight-line one that least squares already solves cleanly.

Reading the model

The model has two coefficients. The first, the multiplier out front, is the value of y when x is zero, the starting level of the curve. The second, the base raised to the power x, is the factor by which y multiplies for each one-unit step in x.

That base is the key to the behaviour. If it is greater than 1, the curve grows, and the further above 1, the faster. If it is between 0 and 1, the curve decays toward zero. A base of 2 means y doubles with every step in x, while a base of 0.5 means it halves. Reading the base tells you at a glance whether you are looking at growth or decay, and how fast.

A worked example

Suppose you enter the points (0, 2), (1, 4), (2, 8), and (3, 16), where y doubles each time x rises by one.

The calculator returns the model y = 2 · 2^x: a starting value of 2 and a base of 2. The base of 2 confirms the doubling, and the multiplier of 2 is the value at x equals zero. To predict y at x equals 4, put it in: 2 times 2 to the power 4 = 2 times 16 = 32, the next double in the sequence.

Entering your data

Enter at least three x,y pairs, separated by spaces or new lines, each pair written as two numbers with a comma between. The y values must be positive, since the fit relies on their logarithms. The calculator returns the two coefficients, the model equation, and a plot of the curve against your points.

Questions people ask

What is exponential regression?

A method for fitting a curve of the form y equals a times b to the power x to data that grows or shrinks by a constant factor, rather than a constant amount.

When should I use it instead of a line?

When y multiplies by a roughly constant factor for each step in x, such as doubling or halving. A line suits data that changes by a constant amount instead.

What does the base tell me?

The factor y multiplies by for each one-unit step in x. Above 1 means growth, between 0 and 1 means decay, and a base of 2 means doubling.

Why must the y values be positive?

Because the method fits a line to the logarithm of y, and logarithms are only defined for positive numbers. Zero or negative y values cannot be used.

References

A quick note on where the methods here come from. Fitting exponential models and the use of transformations are set out in the NIST/SEMATECH e-Handbook of Statistical Methods, the US government's public statistics reference. OpenStax Introductory Statistics is a free, widely used textbook covering regression and least squares.

1. NIST/SEMATECH e-Handbook of Statistical Methods (process modeling and transformations). https://www.itl.nist.gov/div898/handbook/
2. OpenStax, Introductory Statistics (regression and least squares). https://openstax.org/details/books/introductory-statistics-2e