Data is everywhere. In fact, the amount of digital data that exists is growing at a rapid rate, doubling every two years and changing the way we live. According to IBM, 2.5 billion gigabytes of data was generated every day in 2012. Another article by Forbes states that data is growing faster than ever before, and by the year 2020 about 1.5 megabytes of new information would be created every second for every human being on the planet. This data includes: unstructured data, data that is collected from social networks, emails, blogs, tweets, digital images, digital audio, video feeds, censored data, web pages and so on. Semi structured data such as XML files, system log files, text files and so on. Finally, the structured data, data that includes databases, transaction data and other structured data formats. The big data concept refers to the large collection of heterogeneous data from different sources, and it encompasses unstructured, semi-structured, and structured data. Big data processing cannot be easily achieved using traditional data analysis methods. Instead, unstructured data requires specialized data modelling techniques, tools and systems to extract insights and information as needed by organizations. Data science is a scientific approach which applies mathematical and statistical ideas and computer tools for processing big data. Organizations need big data to improve efficiencies, understanding markets, and enhance competitiveness, whereas they also need data scientists that utilize methods or mechanisms to derive insights from big data in a timely manner. Currently, for organizations there is no limit to the amount of valuable data that can be collected, but to use all those data to extract meaningful information for organizational decisions, data science is needed. Therefore, there is an increasing demand for data scientists by companies. Companies use big data and data science for many different purposes. Customer acquisition and retention, offering marketing insights, risk management, product development and increasing efficiency of satellite management are just some of them. It is well known that companies need to listen to customer feedback to innovate. The digital technologies with their increasing ability to store data, open up what we can call the era of feedback. This means that we can get immediate feedback on anything. Because of these new technologies, however, the amount of information that companies have to process is becoming larger and larger. Take the example of a new mobile application. Listening to customer feedback is hard if this feedback is made of millions and millions of reviews. The good news is that modern machine learning techniques that can help us to make sense of all this information. In a research project initiated here at Bocconi, we map the competitive environment of the mobile dating application industry. Define it as all the applications that can help people finding their romantic partners. Developers of these applications face a very practical problem: What are the features that are most relevant for customers? Take the application eHarmony as an example. It focuses on developing the most accurate matching algorithm to help people finding the perfect match. When you open the app, you get many questions about what you like and what you don't like. This information is then used to develop the algorithm. On the other hand, you can take another direction, like Bumble, and focus on the need of a specific category, like women. In our project, we use a topic modelling technique to analyze more than 3 million user-generated reviews on all the applications, and identify the most relevant features that people appreciate. Using these methods, we are able to draw the performance dimension map. Each dimension is related to some words that are frequently used by customers. Once we establish the performance dimensions map, we can understand how the different applications perform along the different dimensions. Take eHarmony as an example, as expected, the app works well in helping people to find their perfect partner. This is represented by many people writing reviews, including phrases like ‘my love’, ‘perfect match’. The app Tinder is the opposite example. Instead of focusing on the complex and time consuming profiling of users, the developers made the app extremely simple, entertaining, and fun to use. In the company's founder words, the interface was designed as a game. This is reflected in their reviews, like the app is a fun game. We can then combine these data with the performance data of these applications, like the number of downloads, to find the strategic insight. It turns out that offering a simple aspect, like having an application that is fun to use, is more relevant for users than having very sophisticated matching algorithms.
