As a transitional step, this site will temporarily be made Read-Only from July 8th until the new community launch. During this time, you can still search and read articles and discussions.

While the community is read-only, if you have questions or issues requiring TIBCO review/response, please access the new TIBCO Community and select "Ask A Question."

You will need to register or log in or register to engage in the new community.

AI on Demand: Data Science in Operations

Last updated:
7:01pm Aug 01, 2022

The TIBCO Data Science team has created "AI on Demand" -- a compilation of innovations to provide guidance, examples, accelerators and starter apps for running AI systems at scale. The program shows real-time AI in action, including case studies in recommender systems, anomaly detection, risk management, dynamic pricing and customer engagement. The resources fall into two categories: AI Apps to provide deep, fully developed applications employing AI to address the world's most challenging use cases, and (2) Tech Tools which are general self-service tools to assist in more ad-hoc data analyses.

All of these offerings are based on real-world customer opportunities. The goal is to provide a brief yet rich set of capabilities that showcase how TIBCO envisions an AI-enabled world.

AI Apps

AI Apps use capabilities across the TIBCO platform to address deep and complex industry challenges. View the gallery below with links to demo videos, blogs, and templates to help you get started.

Machine Learning for Pattern Recognition

| Manufacturing | Image Analysis |

Click on the image above to watch a demo.

Endless streams of data should be our savior, but too often they drown us. Identifying patterns of interest in big data is complicated, but remains a critical part of any manufacturing scenario. Explore TIBCO’s novel approach to pattern recognition as applied to semiconductor wafer maps. Using a combination of machine learning techniques, we have enabled models to sift through massive datasets and present the manufacturing experts with a digestible and visual representation of patterns. The users can then apply their own expertise and power of the human brain to improve the model iteratively. The resulting model is deployed in production environments to act on incoming manufacturing data. While TIBCO Spotfire serves as the powerful visual and interactive user interface TIBCO Data Science - Team Studio unleashes the scale of Spark clusters. TIBCO Data Virtualization serves as the data abstraction layer to surface various data sources in a manageable and user-friendly way. This is a great example of how TIBCO envisions a smart manufacturing architecture. 

Anomaly Detection and Root Cause Analysis

| Manufacturing | Energy | Transportation | Cloud |

The TIBCO anomaly detection solution leverages cutting edge Tensorflow Autoencoders to identify anomalous behavior in multivariate environments. This methodology maps many variables into a lower dimension and compares the predicted value to the actual readings; if the predicted value is far from the actual reading it is labeled as an anomaly. This approach also offers the user visual analyses to find causes of the anomalous behavior. 

Statistical Process Control

| Manufacturing | Energy | Healthcare | Telco | Cloud |

Control Charts have played a vital role in the evolution and success of the global manufacturing industry. Today, SPC is widely used across the globe and considered an essential tool in the mass production of manufactured goods. Its usage is growing in many other sectors, as well, such as Energy, Healthcare, and Telco.
The SPC AI App shows a solution to monitor a large number of parameters leveraging Statistical Process Control methods. The main summary page highlights recently alerted problems requiring attention. Engineers are able to select a parameter, re-calculate actual quality control charts, and interactively investigate the situation.
  • Watch a demo of this application
  • Try it out yourself in the TIBCO Spotfire Demo Gallery
  • Download it from the TIBCO Community Exchange
  • Learn more about it by visiting the Process Control Monitoring and Alerting Solution page.  It provides a more detailed explanation of SPC methodology and this AI App

Dynamic Pricing

| Insurance | Finance | Retail |

Insurance quoting systems are becoming increasingly online and self-service, generating new data rapidly. Insurance brokers sell policies based on premiums priced by insurers. If brokers are able to utilize new data effectively, they may offer more competitive premiums by using their commissions to provide discounts to select customers. Our solutions show how TIBCO Analytics can not only monitor and visualize new quote data in real-time but also continuously "rebase" machine learning models with new data and improve the model automatically over time. 

Customer Engagement and Recommendations

| Retail | Telecommunications |

This demo guides retail sales and marketing decision makers through a step-by-step advanced analytics application. Underpinned by data science workflows for customer segmentation techniques such as RFM and Customer Lifetime Value combined with Market Basket Analysis for Next Best Action product recommendations, this analysis produces actionable consumer insights using intuitive, interactive sliders, filters, and visualizations. While the demo uses specific types of products as example, the same methodoloty can be applied across various customer metrics and recommendation use cases.

  • Learn from this short demo video how TIBCO can help your telecommunications company deliver real-time contextual offers to customers. 
  • View the full demo on this YouTube playlist or learn from this e-book how companies can create smart apps with TIBCO Data Science. See how "Telco X" improves sales of the newest smartphone and what can be done when predictive analytics is infused into critical business processes.
  • Download the Spotfire template for Customer Analytics from the TIBCO Community Exchange and try it our for yourself!

Digital Twins to Improve Yield

| Manufacturing | Energy |

Click on the image above to watch a demo of this solution 

This is a manufacturing demo about digital twins for yield. It's about real-time, continuous analysis of manufacturing equipment sensor and process data at very large scales - up to millions of predictor columns - to understand the causes of semiconductor product yield loss. Digital twins are virtual representations of physical systems. The recent intense interest in them is fueled by the convergence of IoT, machine learning and big data technology directed at the growing volumes of data available from sensors on process equipment. As the process complexity increases, these digital twins are becoming key to efficient operations and high product yields. Part 1 of this demo features Spotfire and shows how the data is visualized to easily view the results. Part 2 features TIBCO Data Science and shows the big data science workflow used to generate the data visualized in Part 1.

Production Surveillance & Condition-Based Maintenance

| Energy | Manufacturing | Transportation |

Click Here for Demo Video

In this example, we take historical data, determine conditions that help us predict failure, deploy those as rules into a real-time data feed, and monitor for potential new failures with streaming data. This "closed-loop" analytics process helps you tighten your operations and increase uptime.

  • Want to try yourself? See our Github repo with a step-by-step walkthrough HERE
  • Or, just interact with a Live streaming version directly on our Demo Gallery HERE

Fraud and Risk Management

| Finance | Insurance | 

Click on the image above to watch a demo

Fraud detection is an enormously important challenge in the financial and insurance industries. On one hand they are very costly and often difficult to catch. On the other hand, flagging legitimate transactions as fraudulent leads to poor user experience and opportunity cost. This demo shows how TIBCO technology helps analysts and financial professionals explore large volumes of their transactions to build hypotheses. From there we see how labeled data can be used to build a supervised model that labels new transactions as legitimate or fraudulent. We also define a measure of oddity where each transaction is measured based on how familiar it looks compared with prior transactions in an unsupervised manner. Last but not least we see how our models can be shared and operationalized in an enterprise environment to label incoming transactions in real time.

Tech Tools

Tech Tools are useful templates, extensions, and approaches put together by our Data Science Team to assist your self-service analytics. Click on the tools below to learn more.