Multivariate Analysis term is used to include all statistics for more than two variables that are simultaneously analyzed.
Multivariate analysis is based upon an underlying probability model known as the Multivariate Normal Distribution (MND). The objective of scientific investigations to which multivariate methods most naturally lend themselves includes.
Objectives of Multivariate Analysis
The following are some basic objectives of multivariate analysis.
- Data reduction or structural simplification
The phenomenon being studied is represented as simply as possible without sacrificing valuable information. It is hoped that this will make interpretation easier. - Sorting and Grouping
Graphs of similar objects or variables are created, based on measured characteristics. Alternatively, rules for classifying objects into well-defined groups may be required. - Investigation of the dependence among variables
The nature of the relationships among variables is of interest. Are all the variables mutually independent or are one or more variables dependent based on observation of the other variables? - Prediction
Relationships between variables must be determined for predicting the values of one or more variables based on observation of the other variables. - Hypothesis Construction and testing
Specific statistical hypotheses, formulated in terms of the parameter of the multivariate population, are tested. This may be done to validate assumptions or to reinforce prior convictions.
Applications: Multivariate analysis is used in various fields:
- Social sciences (understanding factors influencing voting behavior)
- Business (analyzing customer demographics and purchase patterns)
- Finance (evaluating risk factors in investment portfolios)
- Natural sciences (studying the relationships between different environmental variables)
Multivariate Data Sets
We are concerned with analyzing measurements made on several variables or characteristics. These measurements (data) must frequently be arranged and displayed in various ways (graphs, tabular form, etc.). Preliminary concepts underlying these first steps of data organization are
Array
Multivariate data arise whenever an investigator, seeking to understand a social or physical phenomenon, selects a number of variables $p\ge$ of variables or characteristics to record. The values of these variables are all recorded for each distinct item, individual, or experimental unit.
$X_{jk}$ notation is used to indicate the particular value of the kth variable that is observed on the jth item or trial. i.e. $X_{jk}$ measurement of the kth variable on the jth item. So, $n$ measurements on $p$ variables can be displayed as
\[\begin{array}{ccccccc}
. & V_1 & V_2 & \dots & V_k & \dots & V_p \\
Item 1 & x_{11} & x_{12} & \dots & x_{1k} & \dots & x_{1p} \\
Item 2 & x_{21} & x_{22} & \dots & x_{2k} & \dots & x_{2p} \\
\vdots & \vdots & \vdots & \vdots & \vdots & \vdots & \vdots \\
Item j & x_{j1} & x_{j2} & \dots & x_{jk} & \dots & x_{jp} \\
\vdots & \vdots & \vdots & \vdots & \vdots & \vdots & \vdots \\
Item n & x_{n1} & x_{n2} & \dots & x_{nk} & \dots & x_{np} \\
\end{array}\]
These data can be displayed as rectangular arrays $X$ of $n$ rows and $p$ columns
\[X=\begin{pmatrix}
x_{11} & x_{12} & \dots & x_{1k} & \dots & x_{1p} \\
x_{21} & x_{22} & \ddots & x_{2k} & \ddots & x_{2p} \\
\vdots & \vdots & \ddots & \ddots & \vdots & \vdots \\
x_{j1} & x_{j2} & \ddots & x_{jk} & \ddots & x_{jp} \\
\vdots & \vdots & \ddots & \vdots & \ddots & \vdots \\
x_{n1} & x_{n2} & \dots & x_{nk} & \dots & x_{np}
\end{pmatrix}\]
This $X$ array contains the data consisting of all of the observations on all of the variables.
Example: suppose we have data for the number of books sold and the total amount of each sale.
Variable 1 (Sales in Dollars)
\[\begin{array}{ccccc}
Data Values: & 42 & 52 & 48 & 63 \\
Notation: & x_{11} & x_{21} & x_{31} & x_{41}
\end{array}\]
Variable 2 (Number of Books sold)
\[\begin{array}{ccccc}
Data Values: & 4 & 2 & 8 & 3 \\
Notation: & x_{12} & x_{22} & x_{33} & x_{42}
\end{array}\]
The information, available in the multivariate data sets can be assessed by calculating certain summary numbers, known as multivariate analysis: multivariate descriptive statistics such as Arithmetic Mean, Sample Mean (the measure of location), Average of the Squares of the distances of all of the numbers from the mean (variation/spread i.e. Measure of Spread or Variation).
