End-to-End Guide: Building and Deploying Machine Learning Models in SAS Model Manager

[#1]

Introduction

SAS Model Manager is a powerful tool for building, deploying, and managing predictive models. It offers a centralized repository for models, as well as a suite of tools for managing the entire model development lifecycle. Some of the key capabilities of SAS Model Manager include:

1. Model development: SAS Model Manager allows data scientists and analysts to develop predictive models using a wide range of statistical and machine learning techniques. It supports data exploration, data preparation, model development, and model evaluation.
2. Model deployment: Once a model is developed, SAS Model Manager allows users to deploy the model in a variety of ways, including through REST APIs, SAS Micro Analytic Service (MAS), and batch scoring.
3. Model monitoring: SAS Model Manager includes tools for monitoring model performance and detecting changes in model behavior over time. This is critical for ensuring that models remain accurate and reliable as data and business conditions change.
4. Collaboration and governance: SAS Model Manager allows users to collaborate on model development and share models across teams. It also provides robust governance capabilities, including version control, audit trails, and model validation.

Building accurate predictive models is essential for businesses to gain insights from their data and make informed decisions. Accurate models can help organizations identify trends, detect anomalies, predict outcomes, and optimize processes. They can also help organizations improve customer experience, reduce risk, and increase profitability. However, building accurate models can be challenging, as it requires a deep understanding of statistical and machine learning techniques, as well as expertise in data preparation and feature engineering. SAS Model Manager can help organizations overcome these challenges by providing a comprehensive set of tools for model development and deployment, as well as robust governance and monitoring capabilities to ensure that models remain accurate and reliable over time.

Data Preparation

1- Import dataset

Create a new project in SAS Viya for building models by selecting “Build Models” and then choosing “New Project”.

provide a name for the project and then select the dataset from the browsing option.

Choose the Species variable and then select a target for the role.

2- Explore the dataset

Move the Data exploration node by dragging it and then execute the pipeline by clicking on run.

Select the Data exploration node and then click on the Results option.

After exploring the iris dataset, it can be concluded that the dataset contains information on four features of iris flowers: sepal length, sepal width, petal length, and petal width. The dataset is well-balanced and contains 50 observations for each of the three species of iris flowers. Additionally, the dataset is relatively clean and requires minimal data preprocessing.

Model Building

In the supervised learning section, include Forest, Decision Tree, Gradient Boosting, Logistic Regression, and Neural Network models.

Select “Model Comparison” and then choose “Results”.

Based on the model comparison, the Gradient Boosting model is the top performer.

Deployment

Choose the Gradient Boosting model in the pipeline comparison and click on “Register Models”.

Next, select “Manage Models” to navigate to the Model Manager.

Proceed to publish the model to SAS Micro Analytic Service.

Sending Inference Requests

The user can utilize Postman to obtain an access token using the password grant type, which is required to access SAS Viya and send requests to the deployed model.

Make sure to include this access token when sending a request to the model along with the input data.

Conclusion

In conclusion, this blog provided a comprehensive and practical guide on building and deploying machine learning models in SAS Model Manager. It highlighted the step-by-step process, from data preparation to model evaluation and deployment. The blog emphasized the importance of accurate predictive models and showcased the capabilities of SAS Model Manager in streamlining the end-to-end model development lifecycle. By following the guide, readers gained valuable insights into leveraging SAS Model Manager to enhance their machine learning projects, ensuring efficient model creation, evaluation, and deployment. Overall, this blog served as a valuable resource for both beginners and experienced practitioners seeking to harness the power of SAS Model Manager for building and deploying robust machine learning models.