The simplest method of outlier detection is the statistical zscore calculation. This
formula, given as (value mean) / standard deviation, provides a simple and compact
method of measuring outliers. The numerator shifts each point to a zerobased scale, and
the denominator adjusts the distribution to a standard deviation of one. Once the data are
transformed into this standardized scale, generalized statements can be made. In the
author’s experience, outlier scores of 5, 8, or even 12 are often found in real world data.
At times these may be the result of nonnormal distributions, but even in those cases, the
score provides an indicator to potential problems.
The simplest method of outlier detection is the statistical zscore calculation. Thisformula, given as (value mean) / standard deviation, provides a simple and compact method of measuring outliers. The numerator shifts each point to a zerobased scale, and the denominator adjusts the distribution to a standard deviation of one. Once the data are transformed into this standardized scale, generalized statements can be made. In the author’s experience, outlier scores of 5, 8, or even 12 are often found in real world data. At times these may be the result of nonnormal distributions, but even in those cases, the score provides an indicator to potential problems.
การแปล กรุณารอสักครู่..
