Using neural networks to predict MBA student success

College Student Journal, March, 2004 by Bijayananda Naik, Srinivasan Ragothaman

There are several limitations that inhibit neural network models from making highly accurate predictions. The following limitations should be considered while interpreting the neural network results. First, the size of the sample used in this study somewhat limits the generalizability of the results. Second, there is not a single widely accepted theory that would guide the design of network topology. Various decisions such as the number of hidden layers, number of nodes, training tolerance etc. must be made by trial and error. Since training is significantly influenced by the chosen parameters, there could be better neural network solutions that remain unexplored with the same data. Third, training a neural network can be computationally very intensive and learning rate determination is subject to researcher preferences. Fourth, neural networks do not provide explanations for their decisions and this may restrict their use. It should be pointed out that the neural network model described in this paper is not meant to replace the need for professional judgment on the part of MBA directors and deans. Rather, it is meant to complement such judgment and to assist in decision making.

Table 1
Neural Network Results

CLASSIFICATION MATRIX      PREDICTED GROUP

Actual         Percent
Group          Correct     Marginal     Successful     Total

Marginal        80.90         51            12           63
Successful      93.38          8           113          121

TYPE I ERROR: 19.10% *
TYPE II ERROR: 6.62%

* Type I (II) error is defined as the percentage of students
that were classified as Successful (Marginal), while they were
actually Marginal (Successful).

Table 2
Logit Results

CLASSIFICATION MATRIX      PREDICTED GROUP

Actual         Percent
Group          Correct     Marginal     Successful     Total

Marginal         46.03           29             34        63
Successful       86.78           16            105       121

TYPE I ERROR: 53.97%
TYPE II ERROR: 13.22%

Table 3
Probit Results

CLASSIFICATION MATRIX      PREDICTED GROUP

Actual         Percent
Group          Correct     Marginal     Successful     Total

Marginal         46.03           29             34        63
Successful       87.60           15            106       121

TYPE I ERROR: 53.97%
TYPE II ERROR: 12.40%

References

Abedi, J. (1991). Predicting Graduate Academic Success from Undergraduate Academic Performance: A Canonical Correlations Study. Educational and Psychological Measurement, 51, 151-160.

Adams, A.J., and Hancock, T. (200). Work Experience as Predictor of MBA Performance. College Student Journal, 34, 211-216.

Ahmadi, M., Raiszadeh, F., & Helms, M. (1997). An examination of admission criteria for the MBA program: a case study. Education, 177, 540-546.

Arnold, L.R., and Chakravarty, A.K. (1996). Applicant Evaluation in an Executive MBA Program. Journal of Education for Business, 71, 277-283.

Baird, L.L. (1975). Comparative Prediction of First Year Graduate and Professional School Grades in Six Fields. Educational and Psychological Measurement, 35, 941-946.