Response Surface

This package contains activities for generating and visualizing response surfaces

Usage

Add activities included in this package to custom workflows as needed.

Included activities

Response Surface - Train

This script performs data preprocessing, followed by training a machine learning model utilizing two predictor variables and one target variable. It subsequently exports the trained model to the ONNX (Open Neural Network Exchange) format for interoperability and saves the resulting artifact. Depending on the specified configuration, the script can either generate a global model applicable across all entities or create distinct, entity-specific models.

Response Surface - Predict

This script imports a pre-trained ONNX (Open Neural Network Exchange) model and performs inference using two predictor variables as inputs. It leverages Plotly to render a 3D surface plot of the prediction results, converting the plot into a serialized JSON string for storage or further usage. The script is designed to handle both, global and entity-specific models.

Response Surface - Extract Extremum

This script processes the JSON representation of a surface plot to identify and extract the specified extremum value (e.g. minimum or maximum). It retrieves the corresponding input variable values and the associated predicted variable value. These extracted values are then stored in predefined static numeric signals for further computational or analytical use.

Further comments

This package only includes the Python activities. No workflows, schedules or other items are shipped with this package.