Outlier Detection for Data Science: Practical Techniques and Tools Explained

Author(s): Souradip Pal

Originally published on Towards AI.

This member-only story is on us. Upgrade to access all of Medium.

Imagine you’re looking at a large dataset filled with customer purchases. Suddenly, you notice one purchase that’s 10 times the average value! It’s tempting to ignore this, but that oddball number could either be a data entry mistake or represent a unique event. This anomaly is what we call an outlier — a data point that doesn’t fit the general pattern.

Outliers can distort analysis, pull trends in the wrong direction, and lead to incorrect conclusions. But here’s the catch: outliers can also offer valuable insights, like uncovering fraud or identifying hidden trends. In this blog, we’ll explore different ways to detect outliers, understand when to keep or remove them, and cover techniques to treat them effectively.

Before we get into the nitty-gritty of detecting outliers, let’s break down the types:

These occur when you’re dealing with one variable. For example, if you’re analyzing the heights of basketball players and one player is significantly taller than the rest, that’s a univariate outlier.

Below in the example, we use Z-scores to detect outliers in a single-variable dataset. Z-scores indicate how many standard deviations a data point is from the mean.

import numpy as… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI