Decision Optimization + AutoAI = Perfect Blend Available in IBM Cloud Pak for Data

Nerav Doshi
Oct 31, 2019 · 6 min read
Image for post
Image for post
Photo by Mike Kenneally on Unsplash

This article is written in collaboration with Julianna Delua

In the last few weeks, I have attended the data science conferences and seen companies talk about how artificial intelligence (AI) is the future. Yet many companies are struggling to adopt AI due to:

· Major gap between what business wants to achieve and the current AI focus.

· Shortage of skilled data scientists and analysts.

· Businesses cannot build high performing models fast enough with the talent that they have.

At IBM, we believe that AI needs to be built with a purpose — to augment human intelligence. We want to provide a pathway for an enterprise to build and scale AI models with trust and transparency so that businesses can directly benefit from AI investments.

AI is not magic. I have seen companies making decisions based on rules or experience. Some use machine learning(ML) techniques and others use optimization. Common business problems such as workforce planning, pricing and promotion plan, fraud detection, and risk management require a combination of predictive and prescriptive models. The field of machine learning and mathematical programming are now interconnected as optimization problems lie at the heart of most machine learning approaches. To optimize AI outcomes, you need to harness the collective power of predictive and optimization techniques.

Combining Decision Optimization and Intelligent Automation in a Unified Environment

Decision Optimization has been one of the best-kept secrets for top performing enterprises. Just like predictive analytics, optimization is a very old technique, and it was first used in military operations in World War II. Later companies started to use it in many business applications to make business decisions. Organizations have seen millions of dollars in cost savings and return on investment. When they use optimization technology, they gain improved customer service, operation efficiency and speed to decision making along with other benefits. What’s new is that data science teams can solve complex problems using optimization and ML in a single environment. This is why IBM Cloud Pak for Data is one of the most unique enterprise AI offerings that can help you predict and optimize business and exploit the rise of decision intelligence with Watson Studio Premium.

First, we made optimization modeling easier. The Modeling Assistant for decision optimization has simplified the model building process for users. A business user can start building models using the model builder and modeling assistant. The Modeling Assistant uses natural language expressions to build optimization models. There are pre-built templates for some common use cases (eg. resource assignment, scheduling, selection and allocation, and supply and demand planning.)

Image for post
Image for post
Modeling Assistant

You can see an example of the use of the Modeling Assistant and how one can develop the model easily using natural language. It makes it easier for data scientists and analysts to collaborate and build models.

Image for post
Image for post
Optimization templates

Second is the deployment. IBM Cloud Pak for Data helps address several challenges common in putting a model into production. The first challenge is the issue of the model’s compatibility from the point of creation to the point of production. Data scientists today use a variety of different tools. Each new tool and language they use must also meet IT requirements to deploy the model. Models often need to be recoded to meet production environment requirements. Monolithic architectures also strain companies on the options they have to deploy models. Another challenge of model deployment is lack of portability that is common to the legacy analytic systems. We addressed the portability to ensure that we support businesses for the variety of model deployment needs.

And last, when we deploy models we need to make sure that they can scale and meet performance needs. Data used during development is relatively static and at a manageable scale. As the model moves forward to production, it is exposed to larger volumes of data and data transport modes. We also have to support batch, real time or web online deployment type. Models need to be deployed into an application and can be accessed by REST calls at scale.

Image for post
Image for post
Batch deployment in IBM Cloud Pak for Data

Augmenting prescriptive analytics with AutoAI and SPSS

Predictive and prescriptive analytics with automation and productivity tooling in IBM Cloud Pak for Data help businesses fill the talent gap. It empowers data scientists and analysts to build and run models faster and in turn accelerate time to value. Organizations can transform projects that used to take months and weeks into ones that can produce better outcomes in minutes or hours. They can also mitigate the potential of significant model degradation.

IBM has added AutoAI capability to IBM Cloud Pak for Data to automate time-consuming data science tasks and dramatically reduce the time needed to complete data science projects. AutoAI prepares data, selects model, and generated rank model pipelines. AutoAI automates challenging parts of the AI workflow, such as data preparation, feature engineering and hyperparameter optimization, that requires years of experience and brute force to master. These parts of the workflow are not only demanding but also are the expensive bottleneck for business.

IBM SPSS Modeler — A citizen data scientist can quickly build machine learning models using powerful drag and drop data analysis and visualization with SPSS Modeler, a well-known visual data science tool that has been available as part of Watson Studio and Watson Studio Premium for Cloud Pak for Data. SPSS Modeler stream will prepare the input data to train and evaluate a machine learning model. We packaged SPSS Modeler and Decision Optimization together in Watson Studio Premium for IBM Cloud Pak for Data to help a business get more from the data science investment.

IBM Cloud Pak for Data provides a fully integrated data and AI platform that modernizes how businesses collect, organize and analyze data and infuse AI built on Red Hat® OpenShift® and powered by IBM Watson. IBM Decision Optimization, SPSS Modeler and Hadoop Execution Engine are available today as part of IBM Watson Studio Premium for Cloud Pak for Data. Watson Studio Premium is a consumption-based, value-added offering that you need to build and run AI models at scale. AutoAI will be available as part of IBM Cloud Pak™ for Data later in Q4 this year.

You can learn more about automation for IBM Cloud Pak for Data, Auto AI, SPSS Modeler and Decision Optimization. Or join us at our live 3-part Virtual Data Science Camp Fall Edition starting on October 31, 2019. You can view the Summer Edition of this popular 3-part series here. Other IBM Watson Studio webinar playlist is here.

You can join AutoAI code generation beta program. Please click here to signup.To learn more about model ops and devops click here

_________________________________________________________________

Nerav Doshi is a worldwide Digital Technical Engagement lead on Watson Studio Premium and Decision Optimization from IBM. He enjoys building solutions that incorporate business intelligence, predictive and optimization components to solve complex real-world problems.

Please reach out to Nerav for any questions or comments!

IBM Watson

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store