Overview
Feature dimensions refers to the number of degrees of freedom that a sample datapoint has. In the standard case, a data point is represented by a vector
{% \vec{v} \in \mathbb{R}^n %}
where {% n %} is the number of dimensions.
When implemented, the input vector is typically represented as a
column vector
of size {% n %}.
{%
\begin{bmatrix}
a \\
c \\
e \\
\end{bmatrix}
%}
Then number of dimensions is synonomous with the number of
features.
Topics
Depending on the type of machine learning algorithm, and the patterns in the dataset, an analyst may need to change the number of dimensions in the dataset.
- Dimension Reduction - When the number of dimensions of dataset is large compared to the number of datapoints, it is often hard for most analytical techniques to arrive at a robust answer. The solution is often to try to find a way to reduce the number of dimensions.
- Dimension Expansion - in contrast to the technique of dimension reduction, some techniques require dimension expansion, particularly if the training algorithm used requires linear patterns. (see linearing patterns with extra dimensions and kernel trick)