Outliers – Why So Important for Data Analytics?

Print Friendly, PDF & Email

Sponsored Post

Data analytics deals with making observations with various data sets, and trying to make sense of the data. When dealing with very large data sets, automated tools must be used to find patterns and relationships. One of the most important tasks from large data sets is to find an outlier, which is defined as a sample or event that is very inconsistent with the rest of the data set. The observation point or value would be distant from the other observations in the data set. This could be due to a number of reasons, the value could from a measurement where the equipment was not working properly or even from human error.

[clickToTweet tweet=”Outliers are important to data analytics. Check out how Intel DAAL helps.” quote=”Outliers are important to any data analytics problems. Click here to tweet.”]Determining outliers from a data set has applicability in many industries, such as medical diagnosis, sensors (IoT), credit card fraud and intrusion detection systems.  Credit card fraud costs all types of companies billions of dollars per year. By identifying the fraudulent transaction accurately and fast, costs can be reduced for everyone.

While the science behind identifying outliers is quite complex, the performance and accuracy of running algorithms over a data set is important for fraud detection. Fraud detection is a great example of finding outliers, since fraudulent transactions as a percent of all transactions is quite small, very much less than 1%.  From some generally available benchmarks, using a multivariate outlier detection algorithm with the Intel Data Analytics Acceleration Library (DAAL), false positives were kept to a minimum, and the performance was measured in the sub 20 millisecond range.

An abstract concept binary computer information data globe being searched with a magnifying glass for internet fraud

Intel DAAL helps software engineers develop innovative applications that show very high performance. All that an application has to do is link with Intel DAAL libraries which are tuned to take advantage of the latest software and hardware systems. By using highly tuned libraries, developers can create new applications faster, bringing new application software to market faster. Intel DAAL, with its wide range of algorithms built in, enables applications to make better predictions in less time, as well as analyze larger datasets on existing machines. Also, Intel DAAL is always being updated to take advantage of upcoming features in the next generation of hardware, so that an application will run fast when the new hardware is released.

Intel DAAL, together with the latest outlier data analytics solutions can quickly and more accurately detect fraud or anomalies in complex and very large data sets. Determining the data that is important and reliable can assist data scientists in making better predictions in a wide range of industries.


Take advantage of powerful performance libraries that optimize your code and shorten development time.