Skip to main content

Big Data and Bayesian Methods

With the rapid growth in the field of big data, researchers often are required to process massive pools of data that are increasing at high speeds and thus require large storage and prompt decision making analyses.  Because of the speed and volume at which data is being collected and processed, researchers require algorithms and techniques that can quickly and accurately crunch numbers and make fast decisions.  Often, Bayesian methods have been applied to machine learning and data science but there is some concern regarding the precision of such methods as they are considered slow given the speed at which data is being collected.  Indeed, such methods are theory backed and fundamentally crucial in analyzing and predicting trends.  Additionally, the methods are relatively flexible by allowing for missing data and additional dimensions.  However, the efficiency of Bayesian methods is still in question.  Recently, researchers in Tsinghua University, Beijing, China reported current progress in development of flexible, easily scalable algorithms for distributed systems.  They discuss using nonparametric Bayesian models, inferences, algorithms, and systems that are known for their flexibility.  However the issue in regards to speed efficiency lies in how most of these processes still require human involvement, which adds time in the decision making process; with more advances in machine learning these models are believed to work faster and in more robust ways without requiring high time usage.

Bayesian models are fundamental to understanding probability and numerical trends.  Using them in relatively simple problems as discussed in lectures has shown how they can aid in developing conclusions regarding data and trends.  Though we worked with Bayesian probability, it was evident that this probability model could be scaled and used to analyze various parameters.  Because of its flexibility in one area of number analysis, Bayesian methods can be seen as being flexible enough to be applied in other areas, such as the growing area of big data.  Big data is rapidly growing field and its important to find efficient ways of analyzing data without wasting time, storage, and resources.  However its truly interesting to see that a topic so fundamental can be implemented in such a high performance requiring field.



EurekAlert. “Advances in Bayesian Methods for Big Data.” EurekAlert!, AAAS, 31 May 2017,


Leave a Reply

Blogging Calendar

November 2017
« Oct   Dec »