Outlier Detection Toolkit

Compare multiple outlier detection methods on your 2D data

Input Data (CSV Format)

Enter your data with X and Y columns. Each row represents a point.

Detection Methods & Thresholds

Help

How to Use
  1. Enter your 2D data in CSV format with X and Y columns
  2. Or click "Generate Sample Data" for example data
  3. Select which detection methods to use
  4. Adjust sensitivity thresholds for each method
  5. Click "Detect Outliers" to analyze
  6. Review the scatter plot and comparison table
  7. Optionally remove outliers and see the impact
Detection Methods
  • Z-Score: Flags points more than N standard deviations from mean (typically 3)
  • IQR: Uses interquartile range; flags points beyond Q1-1.5×IQR or Q3+1.5×IQR
  • Isolation Forest: Identifies points that are easy to isolate (fewer splits needed)
  • Local Outlier Factor: Detects points with much lower density than neighbors
Understanding Thresholds
  • Z-Score: 3 is standard; 2 is more sensitive, 4 is more conservative
  • IQR: 1.5 is standard; lower values find more outliers
  • Isolation: Higher values are more sensitive
  • LOF: 1.5+ indicates outliers; higher values require stronger evidence
Example Data Format 
X,Y
5,10
7,12
8,15
10,20
50,100
6,11
9,18