Home Documentation Overview Binning Strategies

## Binning Strategies

Below are explanations of the different strategies for Binning (stratifying) Land variables (global variables) when conducting a Representativeness Analysis. For instructions on how to conduct a Representativeness Analysis, click here.

For Continuous Variables:

1. Equal Frequency:  The land variable is divided into bins, each with an equal number (frequency) of GLOBE Land Units (GLU), corresponding to equal areas of Earth’s land.  The minimum and maximum limits of the bins in this strategy will be the minimum and maximum values of the land variable after filtering, so it is possible for collection data to be “out of range”.  This strategy is similar to Stratified Sampling by Proportionate Allocation.
2. Equal Frequency (full range):  Same as Equal Frequency, above, except that minimum and maximum of the land variable are set to the global minimum and maximum of the land variable.
3. Equal Interval: The land variable is divided into bins of equal width, dividing the range of the land variable into equal parts. The minimum and maximum limits of the bins in this strategy will be the minimum and maximum values of the land variable after filtering, so it is possible for collection data to be “out of range”.
4. Equal Interval (full range): Same as Equal Interval, above, except that minimum and maximum of the land variable are set to the global minimum and maximum of the land variable..
5. Diagonally-cut Histogram: A binning strategy that attempts to blend the desirable features of the equal frequency histogram with those of the equal interval histogram. It does so by “diagonally cutting” the empirical cumulative density function.
See: Denby, L., & Mallows, C. (2009). Variations on the Histogram. Journal of Computational and Graphical Statistics, 18(1), 21-31.
6. User Defined:  A binning strategy in which the user can define bin limits used to generate a histogram.
7. GLOBE Legend: Uses the bin definitions defined by the Globe Team to color the raw data map.

For Discrete Variables:

1. Categorical: When land variables are already in a categorical format, (ie Biomes), then these categories are used as the bins (nominal variable).

For either Continuous or Discrete variables:

1. GLOBE Default: Uses “categorical” for categorical (nominal) variables, or “equal frequency, 5 bins” for continuous.