A number of normalization methods can be written as expressions, or used when transforming data. See the links at the end of this topic for a description of the theory behind each method.
In the expression examples below, the following values are used:
Columns: E and A, where E is the column to normalize and A is a baseline column.
Percentile value: P
Normalize by mean
[E] / Avg([E])
[E] * Avg([A]) / Avg([E])
Normalize by trimmed mean
[E] / TrimmedMean([E], P)
[E] * TrimmedMean([A], P) / TrimmedMean([E], P)
Normalize by percentile
[E] / Percentile([E], P)
[E] * Percentile([A], P) / Percentile([E], P)
Scale between 0 and 1
If( Max([E]) = Min([E]), 0.5, ([E] – Min([E]) / (Max([E]) – Min([E])) )
Subtract the mean
[E] – Avg([E])
Subtract the median
[E] – Median([E])
Normalization by signed ratio
If( [E] > [A], [E] / [A], -[A] / [E])
Normalization by log ratio
Log10( [E] / [A] )
Normalization by log ratio in standard deviation units
Log10( [E] / [A] ) / StdDev(Log10( [E] / [A] ))
Z-score calculation
([E] – Avg([E])) / StdDev([E])
Normalize by standard deviation
[E] / StdDev([E])
See also:
Normalization by Log Ratio in Standard Deviation Units
Normalization by Standard Deviation