Impact Plots

Impact Plots explain feature influence on the selected attribute such as cumulative production by calculating the marginal effect that each featue has on the predicted outcome of a machine learning model. These plots are generated using a robust algorithm that determines the optimum engineering parameters for the next well such as lateral length, number of completions, fluids and so on. The output is a number of plots, one for each variable, that shows the optimum engineering parameters. See Production Prediction Workflow for an example on how to determine the optimum engineering parameters in a specific workflow.

Impact Plots using marked data

When you are setting up your machine learning model parameters, you can select to Limit data using markings. However, if you run Impact Plots on this model, the histograms under each chart still reflect all of the data, not just the marked data. The line chart, however, accurately reflect just the marked data.

To adjust the histograms to display only the marked data:

1. Right click on the histogram below the line chart.

2. Select Properties.

3. In the Properties dialog, select Data on the left

4. Under Limit data using markings, check the box for the specific markings.

5. Click Close. The chart should adjust accordingly.

Note that you need to repeat this for each chart. Currently there is no global setting.