Big Data is different. Getting the data used to be the hardest part in research – especially where any kind of human behavior was concerned, such as when we try to understand our customers and users.
But when humans interact with computers, huge amounts of data are produced for free, and the challenge becomes to learn from such unstructured observational data.
To understand Big Data, we need to focus on building “high-luminosity” data microscopes, systems can can turn huge quantity of data into new quality and help us uncover patterns that would otherwise be invisible.
In this talk, I discuss three principles that make Big Data different from classical data analysis, and how clustering and visualization techniques from unsupervised machine learning, with roots in anthropological research long before computers were used, can become a crucial tool for understanding Big Data.
Christoph Best is a data scientist in Google’s Advanced Measurement Technologies research team, where he works on algorithms and tools for data science at scale, with a focus on quantitative marketing science and audience understanding.
Christoph has worked with big data sets all his professional life, starting as a computational high-energy physicist with DESY Hamburg and the MIT Center for Theoretical Physics in Cambridge, Mass., and then as a bioinformatician with a focus on machine learning for structural and bioimage informatics at the Max Planck Institute of Biochemistry in Munich, Germany, and the European Bioinformatics Institute in Cambridge, England.
His particular interests center on the development of Data Science as a discipline and its relation to statistics and machine learning. Christoph holds a Ph.D. in Theoretical Physics from Johann Wolfgang Goethe University in Frankfurt, Germany.