Weighted Regression Prediction Calculator (Model Choice)

Fit weighted (frequency) data with a chosen model and predict y at a given x. Supports linear, quadratic, power, exponential and logarithmic models, with weighted R squared and a scatter plot.

Input

Enter frequency-weighted (x, y) data, pick a model, and predict y at a chosen x. Each line is x, y, frequency, and a missing frequency is treated as 1.

Data points (x, y, frequency)

One point per line. Separate values with a comma or space. A missing frequency becomes 1.

Model

x to predict

Result

Predicted y at x = 6

11.842857

Model equation

y = 0.066 + 1.963 x

Model

Linear y = a + b x

R squared

0.997

Total frequency

Data points

Scatter plot and prediction

How it works

Each data point carries a frequency (weight). The chosen model is fitted by weighted least squares and used to predict y at a given x. The frequency is treated as how many times that point was observed.
Supported models are linear y = a + b x, quadratic y = a + b x + c x squared, power y = a x to the power b, exponential y = a e to the power b x, and logarithmic y = a + b ln x. Power, exponential and logarithmic models are linearized with a log transform before fitting.
The power model needs both x and y positive, the exponential model needs y positive, and the logarithmic model needs x positive. If any point breaks these conditions, that model cannot be computed.
R squared is computed in the original y space, weighted by frequency. A value close to 1 means the model fits the data well.
In the scatter plot the point size encodes the frequency, and the regression curve plus the prediction point at the chosen x are highlighted with a red marker.

Reviews

Tell us what you think of this calculator.