# Stats terminology

version from 2017-06-17 13:43

## Section

QuartilesThe three points that divide the data in ascending order into 4 equal groups which comprises a quater of the data. A quartile is a type of quantile.
QuantileGiven an ECDF p=Fn(x), the quartile is defined as its inverse, Fn-1(x)
PercentileIs a quantile of a particular percent point, Fn-1(p), 0 < p < 100
Individual spaceThe space consists of elements (points) each of which corresponds to each record (row vector) of relational data.
Variate spaceThe space consists of elements (points) each of which corresponds to each variate (Column vector) of relational data.
Orthogonal When the angle between two vectors, xi and xj, is 90 degrees. Thus the inner product = 0.
LevelsThe elements of the domain of factor variables.
The factor variableIs a variable whose data class is character.
The signal noise ratioGiven a linear model, y=Xb + e, the SN ratio is defined as the ratio of squared norms llXbll^2 / llell^2
Degree of freedomThe dimension of the variate space to which the data vector belongs in the linear context.
The random variable A variable which has not got an actual number yet, but will have one which is governed by a probability distribution.
The p-valueThe p-value is the upper (and/or lower) probability of a test statistic of interest under the null hypothesis.