Fit Lines

In a fit line, the data points are fitted to a line that usually does not pass through all of the data points. The fit line represents the trend of the data. Some fits lines are regression based. Others are based on iterative weighted least squares.

Fit lines apply to scatterplots. You can create fit lines for all of the data values on a chart or for the data values in groups, depending on what you select when you create the fit line.

When you add a fit line to a scatterplot, a label is automatically added to the fit line. This label displays the equation used to draw the fit line. You can remove this label or manually add it later in the same way that you hide and show data labels for other elements in the chart. See the topic Data Labels for more information. Note that you cannot show the label on fit lines in scatterplot matrices.

How to Add a Fit Line

  1. If you want to add a fit line for graphic elements in a particular group, select the group. Otherwise, you do not need to select anything in the chart.
  2. If you selected a specific group or want to add a fit line for all groups, from the menus choose:

    Elements > Fit Line at Subgroups

  3. If you want to add a fit line for all graphic elements in the chart, from the menus choose:

    Elements > Fit Line at Total

  4. Use the Fit Line tab to specify the options for the fit line.
  5. If necessary, use the Lines tab to specify the formatting for the fit line.
  6. Click Apply.

How to Edit a Fit Line

  1. Select a fit line.
  2. From the menus choose:

    Edit > Properties

  3. Use the Fit Line tab to specify the options for the fit line.
  4. If necessary, use the Lines tab to specify the formatting for the fit line.
  5. Click Apply.

How to Delete a Fit Line

  1. Select a fit line.
  2. Press Delete.

Using the Fit Line Tab

Display spikes. Draw spikes (connecting lines) from each data point to the corresponding point on the fit line.

Suppress intercept. Do not use a constant intercept value for the fit line calculation. Therefore, the fit line is drawn through the origin.

Mean of y. Draw a line with no slope at the mean of the y-axis data values.

Linear. Draw a regression line with a linear slope that best fits the data points. The regression is computed using the least squares method and a constant (unless Suppress intercept is selected).

Quadratic. Draw a regression line with a quadratic slope that best fits the data points. The regression is computed using the least squares method and a constant (unless Suppress intercept is selected).

Cubic. Draw a regression line with a cubic slope that best fits the data points. The regression is computed using the least squares method and a constant (unless Suppress intercept is selected).

Loess. Draw a fit line using iterative weighted least squares. This method uses the specified proportion of data points to calculate a local smoother. The default proportion is 50%. In addition to changing the proportion, you can select a specific kernel function. The kernel function specifies which data points in relation to the current point receive more weight. The default works well for most data.

uniform. All data receive equal weights.

epanechnikov. Data near the current point receive higher weights than extreme data receive. This function weights extreme points more than the triweight, biweight, and tricube kernels but less than the Gaussian and Cauchy kernels.

biweight. Data far from the current point receive more weight than the triweight kernel allows but less weight than the Epanechnikov kernel permits.

tricube. Data close to the current point receive higher weights than both the Epanechnikov and biweight kernels allow.

triweight. Data close to the current point receive higher weights than any other kernel allows. Extreme cases get very little weight.

gaussian. Weights follow a normal distribution, resulting in higher weighting of extreme cases than the Epanechnikov, biweight, tricube, and triweight kernels.

cauchy. Extreme values receive more weight than the other kernels, with the exception of the uniform kernel, allow.

Confidence intervals. Draw a pair of lines around the regression fit line to illustrate the confidence level percentage that you specify. Choose Mean to draw the intervals based on the predicted mean from the regression line. Choose Individual to draw the intervals based on individual predicted y values from the regression line. Confidence intervals are not available for Loess fit lines because they are not regression-based.