Mining Survey Data

Abstract

Surveys are commonly used as an important data collection tool for empirical research in many applications such as social sciences, marketing and pedagogy. Survey data is becoming one of the major data sources in the era of big data. Conventional statistic tools are utilized to perform survey data analysis. Methods in data mining can extend the capabilities of statistics to explore and discover possible nuggets in massive data. While data mining on general databases has been intensive studied, very few has been done on survey data. Considering the specialities of survey data, this paper describes strategies in mining survey data using computational methods. A novel method for data preparation and dependent pattern mining is presented. Experiments on a real survey dataset were conducted to evaluate the strategies. Results on finding meaningful patterns are reported and discussed.