1. Learn
    2. /
  2. Courses
    3. /
  3. Anomaly Detection in Python
Connected

Exercise

KNN for the first time

You will practice KNN for the first time on a version of the Ansur Body Measurements dataset for females. The female version also contains 95 columns but only 1.9k observations.

The dataset has been loaded as females into the environment.

Instructions

100 XP
  • Import the KNN estimator from the relevant pyod module.
  • Instantiate a KNN() estimator with 0.5% contamination and 20 neighbors as knn.
  • Create a boolean index named is_outlier that returns True when the labels_ of knn return 1.
  • Isolate the outliers from females using is_outlier into outliers.