2.1 Statistics for Numeric Fields

BigML computes the measures below based on all instances of a given numeric field and displays these measures in a histogram, as seen in Figure 2.2 :

• Minimum: the minimum value found in this numeric field.

• Mean: the arithmetic mean of non-missing field values.

• Median: the approximate median of the non-missing values in this numeric field.

• Maximum: the maximum value found in this numeric field.

• Standard deviation: the unbiased sample standard deviation.

• Kurtosis: the sample kurtosis. A measure of ’peakiness’ or heavy tails in the field’s distribution.

• Skewness: the sample skewness. A measure of asymmetry in the field’s distribution.