HIV Data Analysis via Rule Extraction using Rough Sets
T. Tettey, Fulufhelo Vincent Nelwamondo, Tshilidzi Marwala · 2007
The paper presents an analysis HIV data obtained from a survey performed on pregnant women by the Department of Health in South Africa. The HIV data is analysed by formulating a rough set approximation of the six demographic variables analysed. These variables are Race,, Age of Mother, Education, Gravidity, Parity and Age of Father. It is found that of the 4096 possible subsets in the input space, the data only represents 225 of those cases with 130 cases being discernible and 96 cases indiscernible. The rough sets analysis is suggested as a quick way of analysing data and rule extraction over Neuro-fuzzy models when it comes to data driven identification. Comparisons of rule extraction using rough sets and using neuro-fuzzy is conducted and the results are in favour of the rough sets.