Last fall, Reid Karabush, principal consultant of Decision Systems Inc., decided to take IBM up on a challenge to their business partner community to develop a Watson-based solution. This presentation is about how Reid Karabush and Mechie Nkengla, Ph.D. of Data Products LLC., decided to build a machine learning solution in six weeks.
Organizations today struggle with the question of what to do with unstructured “big” data and how it can be leveraged to add value to their business. To date, existing solutions are “siloed” and done out-of-context with existing systems that report structured data (i.e. numbers). The result is an inherent mistrust of insights which stem from the leap-of-faith required to bring meaning from results presented out of context. Using all forms of information, whether it be structured or unstructured, helps organizations better understand their business. All data needs to be managed in such a way that it helps to tell the full story. Journalists describe how to tell a full story by clearly addressing the key components of a story with: Who, What, Where, When and Why. Business systems to-date have focused on Who, What, Where and When; but not Why.
Our solution addresses the question of “Why” something has occurred in context with “What” has occurred. AppX is an example of an emerging class of solutions called “Embedded Analytics”. Embedded analytics is commonly defined as when analytical capabilities such as data management, reporting and visualization are built into other business applications and solutions. AppX takes this concept one step further by embedding advanced analytics – which includes machine learning, content analytics and descriptive statistics – into the very same data management and reporting solutions referenced in this definition of “Embedded Analytics”.
AppX was built using IBM Watson Natural Language Understanding service in combination with open source technologies including Python data science libraries, Django, gunicorn, NGINX, bootstrap, PostgreSQL, and Docker running in the IBM Cloud Kubernetes Service.
The outcome was chosen a top-ten finalist in the 2018 North American Watson Build competition, a soon to be commercial app and, most importantly, first-hand knowledge of what it takes to build a solution in the IBM Cloud.
This presentation is their effort to share that experience with you. If you’re looking to hear what it really takes without biases, then you should attend this event. Be prepared to be surprised. They certainly were!
Reid Karabush is Principal Consultant of Decision Systems Inc., in Northbrook, IL. His expertise includes over 20 years specializing in business performance management and analytics. His experience includes architecting and designing systems for numerous firms in addition to mentoring and training software vendors. He worked within the software product management groups of Comshare and Wang Laboratories, focusing on multi-dimensional databases, analytics, and executive information systems.
Dr. Mechie Nkengla is the Founder/Principal Data Strategist at Data Products LLC, Adjunct Professor at the University of Chicago, Illinois, where she has taught a variety of courses ranging from analytical statistics to advanced data mining and machine learning methodologies. Dr. Nkengla has been designing numerous strategic products based heavily on structured and unstructured large-scale data- intensive garnered intelligence focused on operation optimization. She also has leveraged novel mathematical algorithms, data mining, machine learning and statistical analysis to derive actionable insights from big data.
Prior to founding Data Products LLC, she was a director of data science at the center of excellence at Ernst & Young, where she devised the implementation and industrialization of state-of-the-art analytic systems.
5 pm – 5:30 pm Pizza and networking
5:30 pm – 6:30 pm Presentations and Q&A
6:30 pm – 7 pm More networking