As an MS Business Analytics student at UC Davis, I have been engaged with a micro-mobility startup as part of the practicum for the past six months. My MIP (MSBA Industry Partner) operates dockless electric scooters. As a rapidly growing start-up, it wishes to stabilize its growth and improves its unit economics — making more money from its individual scooter. This calls for increased operational efficiency and improved monitoring of its scooter conditions. And this is where our practicum team comes in. On the journey to achieving such a goal, I, along with my team, have had ample practice in decision-centric analytics.
Our practicum team is tasked with implementing a health dashboard that indicates the scooter’s operating condition. This is a well-defined problem. We need to define health and discover scooter features that indicate health and finally map relevant features to a health condition that indicates when a scooter will break or need repair. With a clear goal in mind, our team delved into the problem with a decision-centric approach.
Identifying and Framing the Problem
In the HBR article, “5 Essential Principles for Understanding Analytics”, the author lists “Identifying and Framing the Analytical Problem” as the foremost principle. Like the author, I also consider framing to be a critical part of a good decision process. As decision analytics is an iterative process, it is important to have a strong understanding of the problem and map out a good plan for the ensuing work. Otherwise, the cost of looping back in the iterative cycle would be high.
My practicum team worked at framing and held many brainstorming sessions to build our analytics framework. Our first step was to define health. We immediately noticed that health was an unobserved status. So, we reframed healthy with unhealthy, since the latter can be decomposed into two aspects: aging and illness. And we interpret these concepts in the context of scooters. Aging refers to the depreciation of scooters from usage, whereas illness can be represented by the damages to the scooters. In the following analysis, we treat these two factors as the crucial factors that affect a scooter’s health and use these as the guidelines for our feature engineering. For example, we gather the scooters’ release date, utilization rate to capture depreciation and repair and replacement to encapsulate damages.
Value of Alternative framing
Because the decisions that drive analysis “may be driven by a hunch or an intuition”, according to the HBR article, it is necessary to develop alternative framings in addition to the original framing. From our research in vehicle health monitoring and predictive maintenance, we find that it is a popular approach to perform predictive maintenance on the component of a vehicle and integrate them into a holistic view of the vehicle. So, in addition to the original plan of developing a health dashboard for the scooters, we proposed to develop dashboards for components. This framework aligns with the analytics strategy of “starting small”, which has its benefits. According to Gallup, the “start small” strategy helps the organization to “find its analytical footing and learn what it needs from a technology, data, management and talent standpoint to drive business impact.”
As we go move forward in the project, we found we made the right precautionary step of preparing alternative frameworks. Early this year, the team visited the company’s scooter service center in San Francisco. During the visit, we observed how technicians charge, repair and maintain scooters and gained first-hand knowledge of the company’s operations. We made one important discovery. We found out that the service center would decommission an entire scooter if its chassis is bent or damaged, which is mostly due to human damage or vandalism. This process involves a great amount of randomness. And it is hard to build predictive modeling for a highly random process. To circumvent this, we decided to pursue our alternative framework, which was to perform predictive analysis on each individual component of the scooter and later aggregate them into a holistic model. In this case, we observe the importance of alternative framing. An alternative framework helps us construct a robust problem-solving approach that avoids the risk of chasing down the wrong lead.
Active Immersion in Stakeholder’s Business
The decision-centric framework also motivates the team’s understanding of the MIP’s business. Because the team was provided a well-defined problem with a clear goal to drive business impact, we were eager to accumulate subject matter expertise (SME) and domain knowledge of the MIP’s business. In the beginning, we conducted research into the micro-mobility industry and performed a SWAT analysis to understand our MIP’s strategy and opportunities. Then, we dug deep into areas of vehicle health and predictive maintenance, of which we had little background before the practicum project. In addition, we also brainstormed and drew business diagrams to understand our MIP’s profit model, growth strategy and its operational process.
I also find the active immersion into our MIP’s business process to be a valuable factor in the decision-centric approach. From the HBR article, the author mentions the importance of a close working relationship between managers and quantitative analysts. It is highly beneficial to have the quantitative analysts work in the relevant business area. Our team was able to achieve the immersion through opportunities like the previously mentioned service center visit. This experience helped us engage with the issue at hand and better understand the business pain points of our MIP. On another note, the decision-centric approach has propelled me to ask “softer questions” that helped me truly engage with the stakeholders in our analytics process. Every time the team makes a major step in the project, I always ask questions like “who is the decision-maker(s) in the process?” and “how will our deliverables be valuable to the stakeholder(s)?” This mindset helps me actively align my understanding with the stakeholders and stay aware and nimble in the analytics process.
Having ample practice in decision-centric analytics, I learned about the importance of framing and developed the mindset to always stay aware of the stakeholders’ pain points and needs. With these experiences, I am confident and ready to launch into an analytics career!
References
Davenport, Thomas H., et al. “5 Essential Principles for Understanding Analytics.” Harvard Business Review, 23 June 2017, hbr.org/2015/10/5-essential-principles-for-understanding-analytics.
Williams, Sean, and Bill Petti. “Data Analytics: Starting Small Is Often a Winning Strategy.” Gallup.com, Gallup, 28 Feb. 2020, www.gallup.com/analytics/237440/data-analytics-starting-small-often-winning-strategy.aspx.
