Segment your entire database with the MultivariateBoxPlotOutlierDetectionAnalysis.xlsm template instead of using a K-Means or Hierarchical clustering algorithm. The current template can segment over 500,000 (r) records across 3 (n) variables of data in less than 4 minutes on a Windows Desktop PC. It uses the theory of the box & whisker plot graph to find outliers in each column of data and then create statistically significant segments. The template can be customized to accommodate more variables (n).
An excellent alternative to the K-Means or Hierarchical clustering algorithms:
- no extra software required, only Desktop Microsoft Excel version
- not sensitive to extremely high or low values, relies on Median to define segments
- no need for software training, just use Excel buttons to run
- guaranteed # of segments (between 0 and n+1) every time
- much easier to interpret the results, no Dendograms required
- find segments much quicker, no multiple runs required
Click the 'Multivariate Box Plot Outlier Analysis!' button after downloading the 'MultivariateBoxPlotOutlierDetectionAnalysis.xlsm' template. Your data must be in the grey-shaded cells provided on the template. Click the 'Start Over' button to begin an analysis from scratch. Great for Data Analysis Toolpak users. You may need to do the following before using this template:
- Enable or disable macros in Microsoft 365 files
- Unblock macros from downloaded files
Notes:
- The box & whisker plot algorithm is a more stable database segmentation solution that relies on the median for identifying segments; K-Means and Hierarchical clustering use the mean (average) to define segments making it sensitive to outliers and anomalies.
