A method has been developed for the analysis of curricula via the statistical analysis of examination results. It is based on the visualization of a set of academic subjects characterized by their correlation matrix obtained on the basis of examination results. The method integrates two methods for data mapping: Sammon's mapping and the self-organizing map (SOM). These are based on different principles, and, therefore, they supplement each other when used jointly. The results of analysis of the correlations of subjects by one of the factor analysis methods-centroid method-are presented for comparison. The main proposition that grounds the method of analysis is that, in most cases, a student gets similar marks in the related subjects. If a student is gifted for the humanities, he will be successful in most of the humanities. Mathematical aptitude yields good marks in mathematical subjects. When analysing the curriculum of studies, we know in advance which subjects are mathematical and which of the humanities. However, there are subjects that cannot be assigned to any well-defined class of subjects. Computer science subjects may serve as an example of such subjects. The proposed method made it possible to evaluate the level of mathematization of different computer science subjects or their nearness to the humanities. The necessary data were the results of examination, only. This level is never quantified, but it may be estimated considering the whole complex of subjects that compose the curriculum. The investigator draws qualitative conclusions about the graphically presented numerical results. The analysis gives us a detailed knowledge of the interaction and similarity of subjects. The knowledge makes a basis both for revising the present curriculum of studies and preparing new ones.
[1]
J. Douglas Carroll,et al.
14 Multidimensional scaling and its applications
,
1982,
Classification, Pattern Recognition and Reduction of Dimensionality.
[2]
Gautam Biswas,et al.
Evaluation of Projection Algorithms
,
1981,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[3]
John A. Hartigan,et al.
Clustering Algorithms
,
1975
.
[4]
Arthur Flexer,et al.
Limitations of Self-organizing Maps for Vector Quantization and Multidimensional Scaling
,
1996,
NIPS.
[5]
Gintautas Dzemyda,et al.
Visualization of a set of parameters characterized by their correlation matrix
,
2001
.
[6]
Jorma Laaksonen,et al.
SOM_PAK: The Self-Organizing Map Program Package
,
1996
.
[7]
Teuvo Kohonen,et al.
Self-Organizing Maps
,
2010
.
[8]
Gintautas Dzemyda,et al.
Comparative Analysis of the Graphical Result Presentation in the SOM Software
,
2002,
Informatica.
[9]
A. E. Maxwell,et al.
Factor Analysis as a Statistical Method.
,
1964
.
[10]
Willem J. Heiser,et al.
13 Theory of multidimensional scaling
,
1982,
Classification, Pattern Recognition and Reduction of Dimensionality.
[11]
M. Davison.
Introduction to Multidimensional Scaling and Its Applications
,
1983
.
[12]
Erkki Oja,et al.
Principal components, minor components, and linear neural networks
,
1992,
Neural Networks.
[13]
John W. Sammon,et al.
A Nonlinear Mapping for Data Structure Analysis
,
1969,
IEEE Transactions on Computers.
[14]
J. Rubner,et al.
A Self-Organizing Network for Principal-Component Analysis
,
1989
.
[15]
Helge J. Ritter,et al.
Neural computation and self-organizing maps - an introduction
,
1992,
Computation and neural systems series.