Topics: Coronavirus | AI | Data Science | Deep Learning | Machine Learning | Python | R | Statistics

KDnuggets Home » News » 2020 » Aug » News, Education » Rapid Python Model Deployment with FICO Xpress Insight ( 20:n33 )

Rapid Python Model Deployment with FICO Xpress Insight


The biggest hurdle in the use of data to create business value, is indeed the ability to operationalize analytics throughout the organization. Xpress Insight is geared to reduce the burden on IT and address their critical requirements while empowering business users to take ownership of decisions and change management.



Sponsored Post.

In a study conducted by Gartner, they found that despite advances in AI and ML initiatives, majority of analytics projects fail primarily due the lack of ability to deploy models at the speed of business. Several other studies point to the lack of engagement from business users in the analytical solution development life cycle being the core issue. From my personal experience with working with clients who are taking advantage of analytics in decision making processes, I have observed similar patterns and can attest to the findings of these studies. The biggest hurdle in creating business value from collected data, is indeed the ability to operationalize analytics throughout the organization.

Operationalizing analytics is a threefold problem. It requires bringing in (1) the business team (problem owners) and (2) Data Science team (analytics subject matters experts or model builders), and most importantly (3) technology that can facilitate the collaboration between the two. A solution developer’s ability to create an interactive user interface with rich visualizations (the application) at the early phase of an analytics project in an agile fashion, rather than relying on IT to build the user interface (application) after completion of model development, is the key in forming collaboration between business and Data Science teams. Enabling this collaboration and taking advantage of the business team’s real-time feedback will also help in adoption of the end solution while ensuring that models are validated and deliver what was intended. Xpress Insight is geared to reduce the burden on IT and address their critical requirements while empowering business users to take ownership of decisions and change management. Here are a few examples from actual client deployments of FICO Xpress Insight in production:

  • Build an initial prototype in 4 hours
  • Iterate through 57 application versions with business users in 1 month
  • Collaborative app with 290+ global active users
  • 1,300+ business scenarios explored per month in one deployed app
  • 2,500+ data elements feeding into each business scenario
  • 37 variations of decision logic accounted for
  • 12 banking optimization applications built and productized in 9 months
  • Reduced scenario run time from 4 days to 30 minutes.

Data Scientists and Operations Researchers can now deploy their Python models into user-friendly, interactive, and scalable apps in FICO Xpress Insight through drag and drop functionality. The models can be of any type—statistical regression, forecasting, machine learning-based, mathematical programming, etc.—and utilize any optimization solver. Users can also leverage the Python package ecosystem, explore data, and execute and compare models and their performance. Xpress Insight is a scalable deployment platform that helps organizations build business applications at the speed of business. It enables collaboration between the data scientist and the line of business user by taking highly complex analytic or optimization models and turning them into simple point and click applications that help make real business decisions. Instead of every Data Scientist repeating the process of building out the common components of an end-user application, Insight has all the core building blocks in a repeatable framework that can be shared with the rest of the Data Science team to reduce development time.

FICO Xpress Insight will help Data Science teams and business stakeholders in several ways:

  • Facilitates collaboration between business users and Data Science team
  • Helps Data Science teams rapidly build scalable applications by drag-and-drop without any html/JavaScript or web development knowledge
  • Enables business users to look at early versions of the end-user application and provide feedback on solution design and development
  • Helps Data Science teams to enable business users to run models, perform simulations, compare scenarios, and visualize outcomes with zero software footprint
  • Reduces development effort through built-in user management and scenario management
  • Provides built-in execution load balancing that Data Science team can take advantage of without extra development time
  • Xpress Insight is IT-independent, not hindered by data limitations and is available for both on-premises and cloud installations.

The latest release of FICO Xpress Insight introduces Insight Python model deployment capability that Data Scientists can take advantage of to easily deploy their Python models with minimal changes to their code. The Python model deployment process is depicted in Figure 1.

Figure

Fig 1. Python Model Deployment with FICO Xpress Insight - Workflow

 

Xpress Insight allows Data Scientists to expose their model data and parameters as they build their models. Model data and parameters can then be made editable in Insight views and end-users can load the default data, change values of model data and parameters within their scenario to perform scenario analysis – all this without changing original data. To elaborate, an optimization or analytical model defines a set of input entities to receive scenario data and a set of result entities that will be populated with the optimization/analytical solution. The solution is the set of values representing the decisions identified by the optimization process or values suggested by analytical models along with other data synthesized from these values. In FICO Xpress Insight applications, a scenario contains the input data and results from the model. Business users will be able to modify/adjust/tweak input parameters to these analytical models to run what-if analysis and to run, clone, and compare what-if analyses across multiple scenarios, as shown in Figure 2.

Figure

Fig 2. Scenario Lifecycle in FICO Xpress Insight

 

To summarize, FICO Xpress Insight powers collaboration in the development and deployment process, linking business users closely with the Data Science team, which enables business users to quickly evaluate, understand, and build trust in solutions that complement the way they think about their jobs.

If you are interested in learning more about FICO Xpress Insight Python capability, please visit FICO Optimization Community Group.

A quick tutorial on the 5 steps Data Scientists need to take to deploy their Python models with Insight is available here.

You can view a demo of FICO Xpress Insight here.


Sign Up

By subscribing you accept KDnuggets Privacy Policy