Model Deployment using Solo_Predictor and Model_Exporter

Course Description

Many applications of chemometrics result in a model with the intent to use it in production or testing environments. Solo_Predictor and Model_Exporter are the two main solutions to deploying models built in PLS_Toolbox or Solo. This course describes these two very different approaches to model deployment and provides practical, hands-on examples of how to use each approach, and the situations where each is more appropriate.

Solo_Predictor is a stand-alone server which permits previously-built Eigenvector models to be used to make predictions on new data supplied by a user. We will describe Solo_Predictor installation and configuration to suit the commonest usage needs, and available client communication methods. Its abilities will be detailed, especially its versatile Solo Scripting language, its web-browser monitoring interface, and troubleshooting issues.

Model_Exporter is an add-on feature to PLS_Toolbox or Solo which enables the export or translation of models into text form based on an interpretable format for use outside of these products. These exported models can be used on other systems and programming language environments (such as C# or Java) by using our provided interpreters, or a user-supplied interpreter, to make predictions on new data.

The course includes hands-on computer time using MATLAB and PLS_Toolbox for participants to understand better the differences between the various options.

Prerequisites

Linear Algebra for Machine Learning and Chemometrics, MATLAB for Machine Learning and Chemometrics, Chemometrics I — PCA, and Chemometrics II – Regression and PLS or equivalent experience.

Course Outline

1.0 Overview of model deployment approaches
1.1 Why Deploy Models?
1.2 Inherent Challenges

2.0 Solo_Predictor
2.1 Overview of Solo_Predictor as a prediction server
2.2 Installation and configuration
2.3 Usage methods (sockets, ActiveX/.NET, wait for file)
2.4 Scripting language
2.5 Field Monitor
2.6 How to troubleshoot problems

3.0 Model_Exporter
3.1 Overview of Model_Exporter and supported methods
3.2 Export formats (MATLAB m-files, XML)
3.3 Usage examples
– MATLAB m-files
– XML interpreters for Java or C#

4.0 Comparison of features and performance
4.1 Situations where Solo_Predictor is preferred
4.2 Situations where Model_Exporter is preferred

5.0 Conclusions