Outliers
- Outliers: refers to objects with characteristic different from most of the other objects
Can detect by
- Summary Statistics
- Box Plots
- Other visualization techniques - Plots
- more advanced approaches like Anomaly Detection
Visualization
Modified Box Plot
Modified box plot
- is a variation of the box plot
- can be used to find outliers
Scatter Plot
Also if we just draw the values, it’ll be possible to see the outliers
Models and Outliers
Some Statistical, ML and DM models are sensible to outliers
- e.g. Linear Regression
- it’s important to detect them and only after build a model