Solve the most complex analytical problems using machine learning and deep learning via a single, integrated, collaborative solution with its own automated modeling API

Screenshot of SAS VDML Pipeline with highlight

SAS for Machine Learning and Deep Learning

Enable everyone to work in the same integrated environment – from data management to model development and deployment.

Easily solve complex analytical problems with automated insights

Automatically generate machine learning and deep learning insights to identify the most common variables across all models, the most important variables selected across models, and assessment results for all models. Natural language generation capabilities are used to create a project summary written in simple language for easy report interpretation. Analytics team members can add project notes to the insights report to facilitate communication and collaboration among team members.
Empower users with language options

Don't know SAS code? No problem. You can embed open source code within an analysis, call open source algorithms within a pipeline, and access those models from a common repository – seamlessly within Model Studio. This facilitates collaboration across your organization, because users can do all of this in their language of choice. You can also take advantage of SAS Deep Learning with Python (DLPy), our open source package on GitHub, to use Python within Jupyter notebooks to access high-level APIs for deep learning functionalities, including computer vision, natural language processing, forecasting and speech processing. DLPy supports the Open Neural Network Exchange (ONNX) for easily moving models between frameworks.
Explore multiple approaches quickly to find the optimal solution

Superior performance from massive parallel processing and the feature-rich building blocks for machine learning pipelines let you explore and compare multiple approaches rapidly. Quickly and easily find the optimal parameter settings for diverse machine learning algorithms – including decision trees, random forests, gradient boosting, neural networks, support vector machines and factorization machines – simply by selecting the option you want. Complex local search optimization routines work hard in the background to efficiently and effectively tune your models. You can also combine unstructured and structured data in integrated machine learning programs for more valuable insights from new data types. And reproducibility in every stage of the analytics life cycle delivers answers and insights you can trust.
Boost the productivity of your analytical teams

Data scientists, business analysts and other analytics professionals get highly accurate results from a single, collaborative environment that supports the entire machine learning pipeline. A variety of users can access and prepare data. Perform exploratory analysis. Build and compare machine learning models. Create score code for implementing predictive models. Execute one-click model deployment. And do it all faster than ever before with our automated modeling API.
Reduce latency between data & decisions

To enhance collaborative understanding, the solution provides all users with business-friendly annotations within each node describing what methods are being run, as well as information about the methods, results and interpretation.
Interpret models using simple language

Standard interpretability reports are available in all modeling nodes, including LIME, ICE, Kernel SHAP, PD heatmaps, etc., with explanations in simple language from embedded natural language generation. Export modeling insights as a PDF report that can be shared outside the data science team.

Customers working smarter with SAS

Key features

A comprehensive visual – and programming – interface supports the end-to-end data mining and machine learning process. Analytics team members of all skill levels are empowered to handle all analytics life cycle tasks in a simple, powerful and automated way.

Data access, preparation & quality

Access, profile, cleanse and transform data using an intuitive interface that provides self-service data preparation capabilities with embedded AI.

Data visualization

Visually explore data, and create and share smart visualizations and interactive reports through a single, self-service interface. Augmented analytics and advanced capabilities accelerate insights and help you uncover stories hidden in your data.

Synthetic data generation

Take advantage of generative adversarial networks (GANs) to generate synthetic data, both image and tabular, for your deep learning models.

Automated insights & interpretability

Automatically generate insights, including summary reports about a project and champion and challenger models. Simple language from embedded natural language generation facilitates report interpretation and reduces the learning curve for business analysts. Share modeling insights via a PDF report.

Bias detection

Assess models for both performance and results bias relative to specified groups.

Cutting-edge machine learning

Take advantage of reinforcement learning – through Fitted Q-Networks, Deep Q-Networks or Actor-Critic – to solve sequential decision-making problems, with support for custom environments.

Decision trees under your control

Interactively adjust the splitting and pruning of decision tree nodes to reflect your business knowledge and enforce regulatory constraints.

Automated feature engineering & modeling

Save time and improve productivity. Automated feature engineering selects the best set of features for modeling by ranking them to indicate their importance in transforming data. Visual pipelines are dynamically generated from your data, yet editable to remain a white box model.

Public API for automated modeling

Take advantage of the public API for automated modeling for end-to-end model development and deployment simply by choosing the automation option. Or use this API to build and deploy your own custom predictive modeling applications. See examples on developer.sas.com.

Deep learning with Python & ONNX support

Python users can access high-level APIs for deep learning functionalities within Jupyter notebooks via the SAS Deep Learning with Python (DLPy) open source package on GitHub. DLPy supports the Open Neural Network Exchange (ONNX) for easily moving models between frameworks. Score new data sets using ONNX models in a variety of environments by taking advantage of Analytic Store (ASTORE).

Easy-to-use analytics

Best practice templates enable a quick, consistent start to building models, ensuring consistency among the analytics team. Analytical capabilities include clustering, different types of regression, random forest, gradient boosting models, support vector machines, natural language processing, topic detection, etc.

Network analytics

Augment data mining and machine learning approaches using a versatile set of network algorithms to explore the structure of networks – social, financial, telco and others – that are explicitly or implicitly part of business data.

Highly scalable in-memory analytical processing

Get concurrent access to data in memory in a secure, multiuser environment. Distributes data and analytical workload operations across nodes – in parallel – multithreaded on each node for very fast speeds.

Computer vision & biomedical imaging

Acquire and analyze images with model deployment on server, edge or mobile. Supports the end-to-end flow for analyzing biomedical images, including annotating images.

Code in your language of choice

Modelers and data scientists can access SAS capabilities from their preferred coding environment – Python, R, Java or Lua – and add the power of SAS to other applications with SAS Viya REST APIs.

Cloud native

SAS Viya's architecture is compact, cloud native and fast, enabling you to make the most of your cloud investment regardless of your cloud provider.

View detailed features list

Explore procedures

View CAS actions

See SAS for Machine Learning and Deep Learning in action

Recommended resources for SAS for Machine Learning and Deep Learning

To browse resources by type, select an option below.

- Select resource type
- 书籍摘录
- 产品简介
- 产品说明书
- 信息图
- 分析报告
- 博客文章
- 客户案例
- 执行简介信息
- 文章
- 案例研究
- 概述小册子
- 电子书
- 白皮书
- 白皮书
- 研究
- 系列
- 网络研讨会
- 行业概述
- 视频
- 解决方案简介
- 访谈