Residual Plot Calculator

Analyse prediction errors, reveal model bias, detect unusual observations, compare diagnostics, and export clear residual plots for dependable machine learning evaluation in seconds online.

Residual analysis results

Interactive diagnostic plot

Interpretation

Observation details

# Label Actual Predicted Residual Absolute Squared Percentage Standardized Studentized Status

Data input

Use commas, spaces, tabs, or new lines.
Use one predicted value for each actual value.
Labels are optional and may contain text.
Expected columns: actual, predicted, label, x, group.

Residual and plot settings

Axis limits and labels

Formula used

Residual = Actual value − Predicted value

Positive residuals show underprediction. Negative residuals show overprediction. Balanced errors should surround zero without visible structure.

How to use

  1. Enter equal actual and predicted value lists.
  2. Add labels, groups, or custom x-values when needed.
  3. Select a residual type and diagnostic view.
  4. Adjust thresholds, trend lines, and display settings.
  5. Calculate, inspect patterns, and export the results.

Example data

ObservationActualPredictedResidual
A109.50.5
B1212.8-0.8
C1514.20.8

Residual plot guidance

Random scatter around zero supports a suitable model form. Curves suggest missing nonlinear relationships. Funnel shapes suggest nonconstant variance and unstable prediction uncertainty.

Large isolated points may indicate data errors or unusual cases. Ordered waves suggest autocorrelation. Persistent positive or negative errors indicate systematic model bias.

Frequently asked questions

What is a residual?

A residual is the difference between actual and predicted values.

What should an ideal residual plot show?

It should show random points distributed evenly around zero.

What does a curved pattern mean?

It often indicates missing nonlinear terms or incorrect model structure.

What does a funnel shape indicate?

It usually suggests heteroscedasticity or changing residual variance.

Why use standardized residuals?

They express errors relative to their estimated standard deviation.

When is a residual an outlier?

Thresholds near two or three standard deviations are common.

Can residuals be positive and negative?

Yes. Their signs show underprediction and overprediction directions.

What is a Q-Q residual plot?

It compares residual quantiles with expected normal distribution quantiles.

What does autocorrelation mean?

Residuals remain related across observation order or time.

Should residual mean equal zero?

It should be near zero for many well-fitted models.

Can I export the analysis?

Yes. Export tables, plots, JSON, CSV, and PDF reports.

Related Calculators

Regression Line PlotPolynomial Regression CurveActual-versus-Predicted PlotResidual Distribution PlotQ-Q Plot of ResidualsScale-Location PlotLeverage PlotCook's Distance PlotPrediction Interval PlotPartial Regression Plot

Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.