Regression Analysis Calculators
This page presents tools that summarise links between variables using fitted lines or curves, providing numerical outputs for trend estimation and prediction tasks.
-
Exponential Regression Calculator
A relationship that grows or declines at a steady proportional rate can be described by an exponential curve shaped by an initial value and a continuous rate of change.
Example use: estimating how the height of a small plant increases each day when the growth rate is roughly proportional to its current height.
Inputs: dataset for the independent variable, dataset for the dependent variable, outlier sensitivity
Outputs: initial value, continuous growth rate, growth factor, percentage change per unit of the independent variable, doubling time, half-life, coefficient of determination, correlation coefficient, exponential model using base e, exponential model using base b, potential outliers detected at index positions
Visual: an exponential curve fitted to the data alongside a display of residuals showing how far each point lies from the curve
-
Lasso Regression Calculator
A linear relationship with a penalty applied to large coefficients can shrink less important contributions towards zero while still estimating an overall trend.
Example use: relating the number of minutes spent practising a skill to the improvement score when some data points may have weaker influence.
Inputs: dataset for the predictors, dataset for the responses, penalty strength, sensitivity
Outputs: intercept, slope, r-squared value, sample size, regression equation, mean of predictors, mean of responses, penalty value, data point information, predicted values, residuals, total sum of squares, residual sum of squares, final coefficient shrinkage
Visual: a fitted line showing the lasso trend with a display of residuals for each data point
-
Linear Regression Calculator
A straight-line relationship between two variables can be summarised by a vertical intercept and a slope that describe the long-term trend.
Example use: relating the number of pages read to the time spent reading on different days.
Inputs: dataset for the predictors, dataset for the responses, outlier sensitivity
Outputs: intercept, slope, r-squared value, correlation coefficient, sample size, mean of predictors, mean of responses, regression equation
Visual: a straight-line fit through the data with a separate display showing residuals
-
Logarithmic Regression Calculator
A relationship that rises quickly at first and then increases more slowly can be described by a logarithmic curve shaped by a growth coefficient and an intercept.
Example use: modelling how the time needed to learn a new skill decreases as more practice sessions are completed.
Inputs: dataset for the predictors, dataset for the responses, outlier sensitivity
Outputs: growth coefficient, intercept, goodness of fit, r-squared value, number of observations, regression equation, potential outliers detected at index positions
Visual: a logarithmic curve fitted to the observed data with a residual display
-
Logistic Regression Calculator
A relationship where the response represents a probability can be modelled by an s-shaped curve that approaches upper and lower limits.
Example use: estimating the chance that someone chooses a particular snack based on how many options are available.
Inputs: predictor values, response values, outlier sensitivity
Outputs: intercept, standard error, coefficient, accuracy, precision, recall, f1 score, mcfadden's r-squared, aic value, result equation
Visual: a smooth probability curve with a residual display showing differences between predicted and observed values
-
Poisson Regression Calculator
A count-based relationship can be described by a model that links the expected number of events to changes in a predictor variable.
Example use: relating the number of times a bird visits a feeder to the number of minutes the feeder is left outside.
Inputs: dataset for the predictor variable, dataset for the dependent count data
Outputs: intercept, slope, log-likelihood, deviance, pearson chi-square, aic value, sample size, regression model equation, predicted values, residuals
Visual: a comparison of observed counts with the fitted poisson curve and a display of residuals
-
Polynomial Regression Calculator
A curved relationship can be modelled by combining several powers of the predictor variable to capture turning points and changes in direction.
Example use: relating the height of a thrown ball to the time since it left the hand.
Inputs: dataset for the predictors, dataset for the responses, polynomial degree, outlier sensitivity
Outputs: calculated coefficients, model statistics, r-squared value, sample size, mean of predictors, mean of responses, regression equation, final coefficients, total sum of squares, residual sum of squares, potential outliers
Visual: a smooth polynomial curve fitted to the data with a residual display
-
Power Regression Calculator
A relationship where the response changes as a power of the predictor can be described by a scaling coefficient and an exponent.
Example use: relating the time needed to tidy a room to the amount of clutter present.
Inputs: dataset for the predictors, dataset for the responses, outlier sensitivity
Outputs: scaling coefficient, exponent, r-squared value, sample size, model equation, outliers, residual analysis
Visual: a power curve fitted to the observed data with residuals shown against the predictor values
-
Quadratic Regression Calculator
A parabolic relationship can be described by a vertical intercept, a linear coefficient, and a quadratic coefficient that together determine the turning point and direction of the curve.
Example use: relating the distance walked to the time taken when the pace changes during the walk.
Inputs: dataset for the predictors, dataset for the responses, outlier sensitivity
Outputs: intercept, linear coefficient, quadratic coefficient, coefficient of determination, vertex, parabola direction, regression equation
Visual: a curved quadratic fit with a residual display
-
Ridge Regression Calculator
A linear relationship with a penalty applied to large coefficients can stabilise estimates when predictors are strongly related to one another.
Example use: relating the number of minutes spent on several small tasks to the total time taken when some tasks overlap in effect.
Inputs: dataset for the predictors, dataset for the responses, regularisation strength, outlier sensitivity
Outputs: intercept, slope, r-squared value, correlation, sample size, mean of predictors, mean of responses, penalty, model equation, statistical outliers detected at positions
Visual: a ridge regression fit shown against the observed data with a residual display comparing predicted values and residuals
Regression Analysis FAQs
Regression analysis summarises how a main variable changes with predictors, producing fitted equations that describe trends and expected values.
They generate coefficients, fit measures and error summaries, offering numerical details needed to assess model behaviour and accuracy.
A linear regression model shows how changes in one variable are associated with consistent increases or decreases in another, represented by a straight-line trend.
Quadratic regression fits data with one turning point, producing a curve that rises then falls or falls then rises around a central position.
Regularisation methods such as Ridge and Lasso reduce coefficient size, limiting instability when predictors overlap or sample sizes are small.