Data Science in the Automotive Industry
Envisioning the future, debates around self-driving cars usually only circle around the specific point of time of general adoption. Will it happen in two, five or ten years? How will regulators handle the issue and how will bureaucratic processes impact the speed of the revolution of the automotive industry? While the upside of self-driving cars in an ideal scenario are crisp and not up for debate, regulators and the public apply high standards of scrutiny to the technology – and much has been written about the ethical challenges, which could arise in the wake of technology taking over vehicles across the globe. Critics found themselves supported by recent events, as a Tesla Model S in self-driving mode was involved in a fatal accident. While aggregate statistics still suggest that self-driving cars will generally increase safety on roads, these incidents reassure general scepticism.
Despite all discussion, the car, and the automotive industry as such, remains a vivid center for the innovative usage of data science. Self-driving cars are powered, to a large degree, by modern advances in machine learning and image recognition technology. The more data about driving behavior, responses to obstacles or similar traffic situations is recorded, the more fine grained the machine learning algorithms can operate. Having acquired large data streams throughout the last years, the technology has reached a sophistication, which makes it almost market ready. While human supervision remains key, self-driving cars are “driven” by self-learning and adapting algorithms, which steer the car through the respective environment. But even besides the prominence of self-driving cars, the entire automotive industry is being re-shaped by data science. Here are just a few examples:
Developing smarter routing in case of road accidents
As long as the self-driving car remains more vision than reality, the human factor in car traffic remains a subject for investigation, and car manufacturers strive to better understand human response to abnormalities in traffic. One of the relevant aspects in this regard is the understanding of traffic congestion mechanisms in case of accidents. Which spots within a city are “hot spots” for accidents? How do drivers react in terms of routing? How long does it usually take for authorities and police to arrive and when is regular traffic back on?
The collection of data through sensors is producing great resources for tackling these questions. Understanding the patterns allows car manufacturers to develop tailored navigation tools, which direct drivers towards optimal routes in case of disturbances. Car manufacturers are not singularly focused on improving the vehicle, but also the behavior of the vehicle with regard to its environment. As part of a “smart city” vision, the development of those real-time routing systems, taking into consideration all dependencies within the system, is a crucial component. Times are gone where customers judge their vehicle simply by style and technical features – rather, focus has shifted to functional benefits like time-saving routing options.
Optimizing user experience while subtly minimizing accident risk
The fact that insurance companies are heavily interested in car and traffic data is long known. Understanding and tracking the behavior of drivers allows them to reward risk-averse driving and offer accordingly low rates, while risky behavior will be penalized with higher rates. Drivers voluntarily opt into this policy (thus, most probably only those who view their responsible driving skills as above average), but certain problems will always remain attached to this issue. Tracking GPS data on individual level could provide the respective company with all sort of other clues about the personal life of the individual, which goes far beyond the initial purpose of data transmission.
At the same time, though, on an aggregated and anonymized level GPS data from cars helps automotive companies to understand certain phenomena better, e.g. the occurrence of speeding or traffic light violations. Being able to detect those instances in the data and subsequently enhancing the understanding of the respective situational context sets the ground for the development of mechanisms, which could intelligently help to prevent them. The vehicle as such does provide various opportunities to enable smarter decision-making of the driver. As long as automotive companies understand the context of potentially dangerous situation in data, they are able to conceptualize tools which not only enhance user / driver experience, but subtly contribute to the minimization of accident risk.
Predicting customer’s purchases
While data science in manufacturing is certainly a core component of the data science impact on automotive companies (see our blog post on manufacturing and data science), there are aspects along the entire value chain, which are experiencing transformation. One of them, is the organization of sales initiatives.
With highly integrated customer relationship management (CRM) systems, this data naturally bears the potential to be turned into valuable insights. When is a certain customer likely to be purchasing again? What after sales activities should be offered to whom? Reliably predicting demand across large customer bases, better segmentation and targeting of sales efforts entails the promise of a better ROI within the sales organization. Instead of using costly and inefficient waterfall tactics, automotive companies can apply more data driven strategies which increase the rate of success. Being more customer-centric also allows them to drive this responsiveness to other functions within the corporation, potentially lifting even more potential.
The car is such a central element in people’s life, that innovation in its industry is moving with rapid pace. Revenue and optimization opportunities arise along the entire value chain and automotive companies face the challenge of adequately prioritizing their roadmap. Support for the core business case can be one goal, while unlocking completely new business cases might also be an option. Balancing risk and reward in an adequate manner, pursuing radical innovation while maintaining focus on the bottom line will be the key distinguisher in the industry for the years to come. As it remains uncertain, when exactly autonomously driving cars will become a reality on our streets, they will be at some point. Nevertheless, while this might be the most visible, it is by far not the only frontier for innovation for automotive companies. Examples for data science applications are manifold and the more success stories emerge, the more will companies be willing to invest time and money to pursue those opportunities.