Regression & Interpolations Guide

Regressions and interpolations are powerful analysis tools that allow you to model data relationships and make predictions based on existing trends. Regressions help create mathematical models by fitting a curve to your data, enabling you to understand relationships and identify key parameters. Interpolations use these regression models to predict values for new data points, bridging gaps in your datasets.

This guide outlines how to create and configure regressions, interpret their results, and use interpolations to calculate new values seamlessly within Benchling.

Note: Regressions, Interpolations, and Analysis Templates require an additional purchase. For access to these features, please contact your Benchling account manager. 

Regressions

Creating a Regression

1. Select Regression when adding a new view under an existing table containing the X and Y variables for the regression.
2. A new regression will be created, and the editing view will open automatically.
3. After creating the regression, you can modify the data from the existing table by filtering or transforming it within the regression view.

Configuring a Regression

The editing sidebar allows you to configure the scatter plot and edit the regression itself:

• Select Axes: Choose appropriate columns for the X and Y axes of the plot.
• Select Regression Type: Choose one of the following:
  • Linear
  • Quadratic
  • 4PL

Optional Configurations

• Weights: Set weights for data points using the "Weights" dropdown. The default is equal weighting for all points. A link to more information is available in the dropdown popover.
• 4PL Constraints: For 4PL regressions, constraints can be set. The default setting is algorithmically determined constraints.

Generating the Regression

1. Click Apply to generate the regression.
2. A regression line will be added to the chart automatically.

Editing a Regression

1. Click Edit to modify the model or weights.
2. Make your changes and click Apply again to update the regression.

Regression Tables

Below the chart, the following tables are available:

• Source Table: Contains the data frame used for modeling.
• Model Output Table: Includes the data used for modeling, predictions, and residual calculations.
• Model Variables Table: Displays parameters resulting from the regression, including:
  • For linear regression: slope and y-intercept.
  • For 4PL regression: Min, Max, Hill Slope, and Inflection Point.
  • Confidence intervals for the parameters are also included.

Interpolations

Adding an Interpolation

1. In the editing sidebar of a view other than the one in which the regression was originally created, scroll to the bottom and click the + next to Transformations.
   • Note: Ensure you are working in a view separate from the one where the regression was originally created.
2. Select Calculate interpolation from the menu.
3. In the modal:
   • Choose the regression model to use.
   • Select the input column from the view containing unknown values.
   • Choose whether the input values should serve as X to predict Y or the reverse.
   • Optionally, select a column to group interpolations by a Series variable.
4. Click Submit.

Results

• A new column named Prediction will appear on the rightmost side of the input table, containing the interpolated values.

• The interpolation step will be represented as a chip in the list of applied transformations and can be modified by clicking the chip.

• Promote the view with the interpolation to a table to add interpolated values to charts or other tables as needed.

If you have additional questions or need further assistance, please reach out to Benchling Support at support@benchling.com.