On Wednesday, November 20th, OPIM Innovate hosted the Hanover Image Analytics Challenge in partnership with Hanover Insurance Group. There, ten teams competed for the opportunity to implement their solutions in a real-world business setting through a spring semester externship. Cash prizes were split between four different thresholds, each having their own requirements. While the lowest threshold provided each participant with a baseline of $100, the highest threshold awarded each team member with $500 and an opportunity to work with Hanover. The objective of the challenge was to have participants make use of a dataset including details on current customer properties and prospective properties to be insured by Hanover.

The difference between this challenge and other case competitions is that it implored students to search for interdisciplinary team members. Rather than having all participants be business majors or intra-major, there were a multitude of participants with little to no business or technical knowledge. As explained by Rob Brewer, Middle Market Vice President and Chief Underwriting Officer at Hanover, the required roles self-assigned by each team member–business analyst, marketing specialist, developer, and data analyst–reflected Hanover’s real-world teams. “Our employees seldom work by themselves,” Brewer said. “They’re always working together across roles.” 

Through the Hanover Challenge, students learned the benefits of cross-boundary teaming. Competitors from OP I AM, one of the Threshold 3 teams recognized during the Awards Dinner following the challenge, made this clear during the Q&A portion of their presentation. When asked what was most challenging while working through their solution, which included the use of data science methodologies to detect and assess certain property holdings, Mariela Kridzelis (MIS & English ‘19), the team’s marketing specialist, commented on the benefits of an interdisciplinary team. “I wouldn’t necessarily go down the challenge route,” she said. “I thought working with my team members was extremely enlightening. We all learned something from each other.” 

Vaughn Nangle, the data analyst of another Tier 3 group entitled Alpha Analytics, also commented on team dynamics. “Throughout the Challenge, my team members and I collaborated so well that we thought we were working beyond our roles. By the very end of the Challenge when we finally presented our solution, that’s when we realized we were actually performing our roles the entire time. In working together, we shared skillsets and expanded each other’s knowledge–it just goes to show how beneficial teams like these are.” The solution Alpha Analytics presented to the judges was a variable heat map of various insurance risk factors. 

We’d like to extend a warm thank you to the Hanover Insurance Group and all faculty and staff that helped with the Hanover Image Analytics Challenge. We’d also like to thank the ten remaining teams of the Image Analytics Challenge for presenting their amazing solutions! 

Please scroll through the slide deck below for team photos.

 

On Thursday, November 14th, OPIM Innovate hosted a Design Thinking workshop in partnership with the UConn American English Language Institute (UCAELI). During this workshop, students participating in the College Preparatory Experience program learned the first three steps of design thinking: define, empathize, and ideate. These steps implore students to define business objectives, understand the needs of their audience, and brainstorm potential ideas for a business solution. Tara Watrous, Innovate’s project coordinator, assisted students throughout the design thinking process by providing an overview of each step. She then provided the following prompt for students to answer in groups: “How can technology be utilized to enhance learning English as a second language?”

Throughout the group portion of the workshop, participating students were able to work with Innovate Lab specialists for insight. As the prompt allowed for many open-ended ideas, each group had their own unique solution at the culmination of the workshop. Ideas included:

• a chatbot
• a virtual reality application where students can immerse themselves in English-speaking environments
• an application for students of different native languages to communicate and practice other tongues
• an augmented reality application that quizzes students on their English by displaying images that represent different dictionary words
• mixed-reality glasses language learners can wear to learn English through environment narration

“The combination of our students’ innovative minds and emerging technology create immense potential for solving problems,” Watrous commented after the event. “Each student group came up with their own unique and creative idea for English-language learning. We hope to see them in our space again to start prototyping!”

 

 

On Thursday, November 7th, Hanover Insurance and OPIM Innovate co-sponsored a Data Visualization workshop. Run by Olivia Rossetti (Management ‘15), Senior Data Integrity Specialist at Travelers and UConn alum, students were introduced to data visualization fundamentals and the Tableau software. 

Rossetti’s framework for introducing students to data visualization began with a general overview of the topic. After defining the purposes of both data and visualization separately, she presented students with the three following definitions: 

In order to visualize data, students had to understand the four key components of data visualization: visual cues, the coordinate system, scale, and context. Whereas visual cues use elements such as shape, size, and color to portray certain values, the coordinate system, scale, and context of a visualization assign additional meaning to those values. Take a bar chart, for example. The length of each bar indicates a numeric value, and color can be used to distinguish between categories. Meanwhile, the chart’s coordinate system can help users determine specific values, and any added labeling, annotations, and titles can help users identify the context of the data being shown. 

Of course, with any chart or visualization, there is always an intended purpose and audience in mind. For data visualization, artists can compare between data points, display parts of a whole, help users identify relationships, or even show the distribution of a dataset. What to do, however, depends on the needs of the audience, including what they may not know and what they could potentially get out of a certain visualization. For example, if someone wants to know the types of users that log onto a website, it wouldn’t be productive to create a visualization showing a distribution of total logins throughout the week. 

After Rossetti walked attendees through the foundational knowledge needed to understand data visualization, she proceeded with the hands-on component of the workshop–working with Tableau. Everyone gained access to the software through the School of Business virtual machine, and exercises were completed in real-time. One of the many visualizations made consisted of using a provided superstore dataset to display sales by product category, customer segment, and market:

Another, more intuitive visualization, consisted of portraying global sales and profit. Size was used to indicate the degree of success for sales. Color was used to indicate profit at a gain or loss:

After the workshop, I approached Rossetti regarding her motivations for returning to UConn, not only as a guest lecturer, but also as an adjunct faculty member. Outside of this Data Visualization workshop, she teaches OPIM 3804: Data Visualization, open to all Management Information Systems (MIS) students and Data Analytics minors. “When I was a student at UConn,” she said. “I wasn’t sure in which direction I wanted to go in. I started as a Management major, finished all of my requirements, and then heard about the certificate in OPIM [the Data Analytics minor]. That led me into the world of analytics, a path I never imagined. Because I had that opportunity as a student, I was exposed to classes that were taught by adjunct faculty. There, I was able to see what they were doing in the real world, and what tools they were using. I knew from the moment I met one of my teachers [adjunct Marshall Dougherty], that I wanted to provide the same experience down the road–I wanted to make an impact.”

We thank Olivia for her contribution to the Hanover Challenge and to the MIS curriculum. We also extend a warm thank you to everyone who attended the Data Visualization workshop!

On Thursday, October 31st, Hanover Insurance and OPIM Innovate co-sponsored a Predictive Analytics workshop. Run by alumni and Travelers Data Integrity Specialist Aleya Hafez (Mathematics & Statistics‘18), students were introduced to different data modeling approaches. 

Hafez’s lecture followed the usual data modeling framework–cleaning data with hypotheses in mind, splitting data, and finally creating predictive models. This, of course, came with the need to explain what each segment entailed, information she generously opened with. In explaining the data cleaning process, she informed students of potential outliers, duplicate records, data input errors, or any other abnormal or extraneous data that may skew test results. With regard to splitting the data, she taught students the difference between a training dataset and a test (or validation) dataset. In brief, training datasets and test datasets are samples taken from the whole; training datasets are used to build initial models, whereas test datasets are there to verify results. As for creating models, Hafez specified the process to further increase understanding.

To provide students with a conceptual guide of data modeling, Hafez broke down the process into the following steps:

From these steps, she then explained separate modeling types, such as linear regression, logistic regression, and decision trees. Whereas linear regression predicts specific numeric values, logistic regression and decision trees predict the probability of something being true or not. For example, a model using logistic regression will be able to predict whether an email is spam or not. A decision tree, on the other hand, brings these probabilities into finer and finer levels depending on the number of nodes, or categories, added (see below):

Decision trees may also be used to assess model outcomes against actual data. This helps data scientists select the best models to use. 

To end the lecture, Hafez used the concepts she described to create predictive models in R. This gave students the chance to see data modeling in action, and provided them with enough of a preview to ask clarifying questions. 

When Hafez was asked why she volunteered her time to share her knowledge on predictive analytics with UConn students, she said the following: “I’ve always enjoyed helping other people, and it excites me to know that someone other than me is interested in statistics. It’s great to see that the UConn School of Business is putting more of an emphasis on predictive analytics, so I’m happy to give my take!”

We thank Hafez for her time and everyone who attended the Predictive Analytics workshop!