Actual-versus-Predicted Plot Calculator

Data input

Example preset

Upload CSV file

Files remain in your browser. Maximum size is 2 MB.

Delimiter

Decimal separator

Header handling

Missing rows

Dataset

Recommended columns: Label, optional Group, Actual, then one or more prediction models.

Model, plot, and reference settings

Active model

Plot type

Model display

Chart title

Chart subtitle

Axis scale

X-axis label

Y-axis label

Decimal precision

Point size

Point opacity

Marker style

Custom line slope

Custom line intercept

Predictor count for adjusted R²

Perfect prediction line

Regression trendline

Confidence interval

Prediction interval

Custom reference line

Mean lines

Equal axis scaling

Gridlines

Legend

Tolerance, outliers, and filters

Tolerance mode

Tolerance value

Upper tolerance

Lower tolerance

Outlier method

Outlier threshold

Metric calculation

Prediction direction

Minimum actual value

Maximum actual value

Minimum residual

Maximum residual

Show top errors

Use zero to display every matching row.

Search observation label

Group filter

Show tolerance bands

Hide detected outliers

Scientific notation

Interactive plot

Click a plotted point to inspect its record.

Observation error table

Rows per page

Formula used

Residual = Actual − Predicted
Absolute Error = |Actual − Predicted|
Squared Error = (Actual − Predicted)²
Percentage Error = ((Actual − Predicted) / Actual) × 100
MAE = Σ|Actual − Predicted| / n
MSE = Σ(Actual − Predicted)² / n
RMSE = √MSE
R² = 1 − [Σ(Actual − Predicted)² / Σ(Actual − Mean Actual)²]
Adjusted R² = 1 − (1 − R²)(n − 1) / (n − p − 1)

How to use this calculator

Paste rows or upload a CSV dataset.
Keep one actual column and one prediction column.
Add more prediction columns for model comparisons.
Choose filters, tolerances, and outlier settings.
Select a plot type and calculate results.
Inspect points, tables, metrics, and warnings.
Export the chart, data, or PDF report.

Example data structure

Label	Group	Actual	Linear Model	Tree Model
H1	Urban	185000	179000	188500
H2	Urban	220000	228000	219500
H3	Suburban	265000	254000	269000

Interpretation guide

Points near the diagonal indicate accurate predictions. Points above it show overprediction. Points below it show underprediction.

Small residuals usually indicate better local accuracy. Systematic residual patterns reveal model bias. Large isolated errors may indicate outliers.

Use several metrics before selecting a model. R² alone can mislead. Error scale and business tolerance also matter.

Frequently asked questions

What does the diagonal line represent?

It represents perfect predictions where predicted values equal actual values.

What does a point above the line mean?

The model predicted a value greater than the observed value.

What does a point below the line mean?

The model predicted a value smaller than the observed value.

Which metric should I prioritize?

Choose metrics matching your error costs, units, and modeling objective.

Can I compare several models?

Yes. Add one prediction column for every model being compared.

How are zero actual values handled?

Percentage metrics omit zero actual values and display a warning.

Can outliers be excluded?

Yes. Select a detection method and exclude marked observations.

What is prediction bias?

It is the average predicted value minus the actual value.

Why use equal axis scaling?

Equal scaling prevents visual distortion around the perfect prediction line.

Is uploaded data sent to a server?

No. This calculator processes uploaded CSV data inside your browser.