Discriminant analysis of bronchitis cancer data

Abstract EN

The aim of this research is to predict membership in two mutually exclusive groups of bronchitis cancer patients, and allocating new patients using Discriminant Analysis.

Discrimination is a multivariate technique concerned with separating distinct sets of objects (or observations) and with allocating new objects (observation) to previously defined groups.

The results showed 90 %, and 98 % of dead and alive patients were classified correctly.

Only 2 % and 10 % of dead and alive patients were misclassified population models.

We now assume that f1(x) and f2(x) are multivariate normal densities, the first with mean vector μ1and covariance matrix 1 and the second with mean vector μ2and covariance matrix 2.

The special case of equal covariance matrices leads to a particularly simple linear classification statistic.