Outliers and Where to Find Them
Overview of Different Outlier Types through an Example
Have you ever wondered why a basic analysis, such as a comparison of monthly average sales, yields a completely different outcome than what is achieved through very depth study? Have you ever wondered how military surveillance detects unusual behavior so quickly, or how your credit card sends you a notification if you make an unexpected purchase?
In today’s environment, data holds tremendous importance. They’re being recorded and analyzed all over the place in the hopes of extracting important data. Every day, new and improved analytical methods and instruments are developed. One of the disadvantages of analysis is that data isn’t always accurate. They may have discrepancies that threaten the entire process. One type of discrepancy is existence of outliers in data.
In simpler terms, outliers are datapoints that is very different from the existing data. They can either be one bad datapoint that disrupts analysis or an unique data point that leads to interesting findings leading to actionable insights.
So, let’s look at the various forms of outliers through an example:
Example
Assume we live in a small community with ten families. These households are mostly made up of retirees who begin their day at 8 a.m. and end it at 8 p.m.
Case 1 : If you observe someone going down the street at 11 p.m. in this neighbourhood, it’s an outlier, and it could be a thief.
Let’s say , a new custom has emerged, in which the entire town gathers on Saturday and throws parties until late at night.
Case 2: If you see someone walking down the street at 11 p.m. on a Saturday in this neighborhood, it is normal; however, on another day, it is unusual. It’s an anomaly, and it could be a thief on that specific context.
With so many elderly people living in close quarters, it’s natural to see someone fall ill from time to time.
