Exponential Regression Calculator
Fit an exponential model y = a × e^(bx) to your data using logarithmic transformation. Calculate coefficients, R², and make predictions. See also our Linear Regression Calculator, Exponential Growth Calculator, and Logarithm Calculator.
How to Use the Exponential Regression Calculator
Exponential regression fits a curve of the form y = a × e^(bx) to your data, where a is the initial value (y-intercept when x=0) and b is the growth or decay rate. When b is positive, the function models exponential growth; when b is negative, it models exponential decay. This model is appropriate when data increases or decreases at a rate proportional to its current value.
Enter your data points as x,y pairs. All y values must be positive since the exponential function is always positive. The calculator uses logarithmic linearization: by taking ln(y), the exponential model becomes linear (ln(y) = ln(a) + bx), allowing ordinary least squares to find the best-fit parameters. Click Calculate to see the fitted equation, R², and predictions.
Common applications include population growth modeling, radioactive decay, compound interest, bacterial growth curves, drug concentration over time, and technology adoption curves. The doubling time (for growth) or half-life (for decay) is automatically calculated as ln(2)/|b|, giving you a practical interpretation of the growth or decay rate.
When comparing exponential regression to linear regression, look at the R² values and residual patterns. If your data curves upward or downward and the linear R² is low, exponential regression may provide a much better fit. However, be cautious about extrapolation — exponential models grow without bound, which is unrealistic for most real-world phenomena over long time horizons.
Formula
Exponential Model:
y = a × e^(bx)
Linearization (take natural log):
ln(y) = ln(a) + bx
Solve using linear regression on (x, ln(y)):
b = (nΣx·ln(y) - Σx·Σln(y)) / (nΣx² - (Σx)²)
ln(a) = (Σln(y) - b·Σx) / n
a = e^(ln(a))
Doubling Time / Half-Life:
t_double = ln(2) / b (when b > 0)
t_half = ln(2) / |b| (when b < 0)
Example Calculation
Data: (1, 2.7), (2, 7.4), (3, 20.1), (4, 54.6), (5, 148.4)
ln(y): 0.993, 2.001, 3.001, 3.999, 5.000
Linear regression on (x, ln(y)):
b ≈ 1.0 (growth rate)
ln(a) ≈ 0.0, so a ≈ 1.0
Equation: y = 1.0 × e^(1.0x) ≈ e^x
Doubling time = ln(2)/1.0 = 0.693 units
Prediction at x=6: y = e^6 ≈ 403.4
R² ≈ 1.000 (excellent fit)
Reference Table
| Scenario | a (initial) | b (rate) | Doubling/Half-Life |
|---|---|---|---|
| Bacterial Growth | 100 | 0.693 | 1 hour |
| Radioactive Decay | 1000 | -0.087 | 8 days |
| Population Growth | 1000 | 0.020 | 35 years |
| Compound Interest (5%) | 1000 | 0.049 | 14.2 years |
| Moore's Law | 2300 | 0.347 | 2 years |
| Viral Spread | 1 | 0.231 | 3 days |
| Cooling (Newton) | 80 | -0.050 | 13.9 min |
| Drug Elimination | 500 | -0.139 | 5 hours |
Frequently Asked Questions
What is exponential regression?
Exponential regression fits a curve of the form y = a × e^(bx) to data where the rate of change is proportional to the current value. Unlike linear regression which fits a straight line, exponential regression captures growth or decay patterns. It is used when data shows accelerating increase (b > 0) or decelerating decrease (b < 0). The method uses logarithmic transformation to linearize the problem, then applies least squares to find optimal parameters.
Why must y values be positive?
The exponential function e^(bx) is always positive, so the model y = a × e^(bx) can only produce positive values (assuming a > 0). The linearization method requires taking ln(y), which is undefined for zero or negative values. If your data contains negative values, exponential regression is not appropriate — consider polynomial regression or a shifted exponential model y = a × e^(bx) + c instead.
How do I choose between linear and exponential regression?
Plot your data first. If points follow a straight line, use linear regression. If they curve upward (accelerating) or downward (decelerating toward zero), exponential may be better. Compare R² values from both models — higher R² indicates better fit. Also check residual plots: if linear regression residuals show a curved pattern, the relationship is likely non-linear. Exponential growth is characterized by a constant percentage change per unit of x.
What is the doubling time?
Doubling time is the time required for the quantity to double, calculated as t = ln(2)/b ≈ 0.693/b. For example, if b = 0.1 per year, the doubling time is about 6.93 years. This is equivalent to the "Rule of 70" in finance: doubling time ≈ 70/(percentage growth rate). For decay (b < 0), the analogous concept is half-life: t_half = ln(2)/|b|, the time for the quantity to decrease by half.
What are limitations of exponential regression?
Exponential models assume unlimited growth or complete decay to zero, which is unrealistic for most real-world phenomena. Populations face carrying capacity limits (use logistic model instead), chemical reactions reach equilibrium, and markets saturate. The linearization method also gives more weight to larger y values, potentially biasing results. For data that levels off, consider logistic regression or polynomial models.
Can I use this for decay curves?
Yes. When b is negative, the model describes exponential decay: y decreases toward zero as x increases. Common applications include radioactive decay (half-life calculations), drug elimination from the body, cooling curves (Newton's law of cooling), and depreciation. The half-life is ln(2)/|b|. Enter your time-series data with decreasing y values, and the calculator will automatically detect the negative growth rate.