A Predictive Modeling graph is one of several data science charts used for in-depth analysis in Domo. It is essentially a Scatter Plot graph that includes a model fit line. You can also specify upper and lower bounds if you want.
The following image points out the primary components of a Predictive Modeling graph:
This chart type does not currently include algorithms to identify your model fit line and upper and lower bounds. You must identify these elements yourself in the DataSet you use to power the graph. However, you can use tools such as R and Python to help you identify these elements.
Powering Predictive Modeling graphs
Predictive Modeling graphs require three columns or rows of data from your DataSet, or five if you plan to show upper and lower bounds. One column contains the X coordinate values for the chart. All remaining columns contain Y coordinate values to be paired with their corresponding X coordinate value to form various elements in the graph. These Y coordinate columns are as follows:
A column with Y coordinate values for the points in the graph (required).
A column with Y coordinate values for the model fit line (required).
A column with Y coordinate values for the upper bound (optional).
A column with Y coordinate values for the lower bound (optional).
For an example of how to set up your data, see the graphic below.
For information about value, category, and series data, see Understanding Chart Data.
In the Analyzer, you choose the columns containing the data for your Predictive Modeling graph. For more information about choosing data columns, see Applying DataSet Columns to Your Chart.
For more information about formatting charts in the Analyzer, see KPI Card Building Part 2: The Analyzer.
The following graphic shows you how the data from a typical column-based spreadsheet is converted into a Predictive Modeling graph. Note that the spreadsheet in this example includes only a small subset of the data used to power the graph.
Customizing Predictive Modeling graphs
You can customize the appearance of an Outliers graph by editing its Chart Properties. For information about all chart properties, see Chart Properties. Unique properties of Forecasting graphs include the following. You can click a thumbnail image to see a larger image.