*by Dedu, Vasile; Armeanu, Daniel
and Enciu, Adrian *

Published in Romanian Journal of Economic Forecasting,
2009, volume 12 issue 4, 170-179

Requires a PDF viewer such as Xpdf
or Adobe
Acrobat Reader 658Kb |

In the present financial theory, we confront with complex economic phenomena and activities which cannot be studied or analyzed profoundly because of the plurality of existing variables, ratios and information. The economic, financial and social activity carried on under crisis or economic growth conditions registered year by year a development of the products and instruments in use. The complexity of the economic area may be simplified through techniques of multi-dimensional analysis. Such a method is the analysis of the principal components which allows the decreasing of the initial causal space dimension generated by the functional links which are established among the initial explanatory variables. The dimension of this space is determined by the number of explanatory variables identified as causes of the economic phenomenon and the higher their number, the more difficult it is to analyze the initial causal space because the information volume, the complexity of calculations, the risk not to identify the contribution of each variable to the creation of the initial causal space variability and the decrease in the initial variables significance in case they would be inter-correlated grow. The simplification of the initial causal space means the determination of a change which consists in transition from a space with a large number of variables to another one of fewer dimensions, equivalent but on the conditions of keeping maximum information from the initial space and maximizing the variability of the new space (called principal space). Variables from the principal space represent the principal components, they are un-correlated and the vectors which define them have a unitary length.

**Keywords**:
original variables, covariance matrix, eigenvalue, eigenvector, principal components, total variance,
generalized variance, factor matrix, factor loadings, factor scores, classification

**JEL Classification**:
G15, G22, C44