Data science is literally everywhere. Many firms look for skilled personnel to cover the subject in their company even if there is very little understanding for the subject. I see a lot of job offers, and talk to a lot of people, which reveals that most projects are not even properly defined, let alone backed by the IT or tier 1 managers. Additionally, managers mistake data warehouse specialist and data scientist, or they are looking for a Swiss army knife. Is this the road to go? Many firms set up projects which cannot be properly supported or are extremely bad defined. I doubt that even half the projects just started meet the business expectations.
This post loosely sums up some use-cases me and some colleagues from different industries identified. So, this is mostly derived from talking to other people’s problems and should show which kind of topics a data science division can support.
- After almost all online merchants adjust prices based on daytime, devices or season, the over-the-counter firms start to put electronic labels on their shelfs and cloths. Using simple decision trees helps big German fashion firms to sell the last seasons goods with as much profit as possible. This trend from the US will come to Europe very quickly. The dependent variable is the number of sales they make setting different prices for similar goods. The routine recommends a price based on the left over stock. Prices are changed over night. The labels receive their price from a wifi network installed in the shop.
- Using text analysis to gain customer insights, e.g. summarize product reviews or complaints management. I firstly noticed it at Google Maps, where reviews were automatically summarized, other firms such as Amazon and Otto (biggest German online retailer) followed. Another example is the automatic analysis of customer complaints to identify the main reasons for trouble.
- Route optimizations for fleed-based services. The potential is huge, if you think about the whole transportation sector, train services or airways. The Deutsche Bahn (the German train company) massively invested in Data Scientists and looks for solutions to reduce delays.
- Using speech-to-text to reduce the cost of bureaucratic barriers. Documentation requirements raise in different sectors like insurance or banking, on a big scale, speech-to-text has huge advantages in cost efficiency compared to humans.
- Online retailers use convolutional neural nets to display similar images on a page to improve the user experience.
- Optimize the internal search function. Most companies have a really messy and badly indexed knowledge data base. Finding important answers saves lives.
- Use scorings to optimize your restricted budget: Contacting the right customer with the right ad. Even simple scoring methods help to reduce the leakage and increases conversion rates.
- Monitor all in-house activities like your employers’ performance (see Amazon).
- Stockholm reduced air pollution optimizing speed limits and increased public transport, using many sensor data and optimizing the whole transportation system. Other cities like Hamburg are also interested in this kind of stuff.
- The most basic idea comes at the end. Base your management decisions on hard facts, meaning a good business intelligence department is a must for bigger firms. They provide regular reports and analysis to evaluate the real business value of your decisions and actions. This is not as common as most people think…
All the points above require the availability of data. Meaning digitalization and data science go hand in hand in many use-cases. Data science can therefore be a chance to redefine the value of digitalization. A second insight from the examples above is, the fact that data science is not only estimating classifiers, there are many optimization routines which can be used when data are available.