Big Data: Heterogeneous Methods for Heterogeneous Data?



Anastasia Ushakova


Teaching Fellow (Postdoc Tutor) - University of Edinburgh

Abstract: The fact that we are living in the age of data is not new to anyone. However, rarely do we come across a complete and inclusive understanding of what contributes to data heterogeneity and how this impacts data collection in the era of big data. The tendency is to assume that forms of big data are more or less alike: volumous, relational, variable, and exhaustive. While the data may often be not hard to access, how do we collect ‘good’ data? How can we learn data generation process of heterogeneous data better? Should we forget everything we know and start from scratch when approaching new forms of data? And as a consequence, how can we teach these methods given such heterogeneity of data and research puzzles?

In my talk, I will present an example of big data by looking at residential energy consumption that arrives from smart meters. This complex time series data is novel but also highly valuable given current energy suitability goals and strategies both in the UK and Europe. The question I will be asking is how can we generate insights from this data and data of similar forms? How much emphasis should be given to the nature of data, how can we address heterogeneity in such data, and how can we teach to approach such data effectively to the new generation of applied data scientists that can tackle these challenges.

Bio: I am a Teaching Fellow in Statistics at the University of Edinburgh (PPLS). I received my PhD from UCL where my thesis looked at big data analysis with applications to social science research questions. I mainly focussed on using data recorded by smart meters to assess users behavioural patterns. The motivation behind my work is to explore whether there is a potential for big data to inform public policy and decision making in energy sector. I am passionate about both development and application of statistical methods as well as promoting those to new generation of social science researchers.