Page 71 - Anthropometry, Apparel Sizing and Design

100 Anthropometry, Apparel Sizing and Design

4.2.4 Step 4: Anthropometric analysis

The final step of Stage 1 is to analyze the data. The statistical method generally applied
at this stage is the descriptive analysis also known as univariate analysis based on sim-
ple statistics. Categorical and continuous data can be analyzed as follows:

4.2.4.1 Categorical data
The categorical data are analyzed to understand the demographic profile of the sample
population. The first classification to be made often is to divide the population into
gender-based subsets, namely, male and female. Frequency distribution curves are
plotted by quantity and percentage, and results can be illustrated using tables and
bar graphs.

4.2.4.2 Continuous data

Continuous data analysis based on descriptive statistics includes calculation of fre-
quency distributions, range, mean, median, mode, standard deviation, coefficient of
variation, and Pearson correlation coefficients to determine the interrelationships
between the various body dimensions.
The objective of anthropometric analysis is to profile the demographic data and the
continuous data in such a way that the overall patterns of body dimensions are
described and one can distinguish between genders and different age groups for selec-
tion of key dimensions.
The next section deals with Stage 2—the sizing data analysis.
In this stage the objective is to divide the sample population into smaller groups com-
posed of individuals who have similar key body dimensions. The center panel of Fig. 4.1
shows the phases of Stage 2, which consists of four steps (Steps 5–8). The analysis
shown in Stage 2 is only one possible method of determining key dimensions and clus-
tering the sample population. Besides the three methods shown here (PCA, cluster anal-
ysis, and decision tree analysis), other methods like bivariate analysis, neural networks,
and artificial intelligence can also be used (Kim et al., 2018; Doustaneh et al., 2010).
Step 5 is multivariate analysis, the purpose of which is to test the sampling ade-
quacy of the collected data. In Step 6, principal component analysis (PCA) is
employed to reduce all the variables into significant components. In Step 7, cluster
analysis is used to segment the sample subjects into homogenous groups with similar
body shapes and sizes. In Step 8 the decision tree technique can be applied to classify
sample subjects into groups based on profiles and to validate the cluster groups.

4.2.5 Step 5: Multivariate analysis
Prior to applying a PCA, a sampling adequacy test needs to be performed on the data to
confirm the appropriateness of conducting PCA to ensure that the data can be factored
well (Tabachnick and Fidell, 2007). In addition, Bartlett’s test of sphericity can also be
used to add a significant value to support the factorability of the correlation matrix
obtained from the items.