Multivariate Adaptive Regression Spline (MARS) Approach for Classification of Sexually Transmitted Infections (IMS) Diseases on Commercial Sex Workers (PSK) In Moroseneng Localization

摘要

Moroseneng localization is a second favorite places of prostitution in Surabaya after Dolly, which located in Benowo District, Surabaya. This place is a source of transmission Sexually Transmitted Infections (IMS) because there are sex workers who is estimated has risk of IMS from their partner. Prevention transmission of IMS can be inhibited on Commercial Sex Workers (PSK) if it had been known the factors that caused to transmission of IMS. The factors which estimated affect the transmission of IMS are PSK's age, region of origin, marital status, education level, IMS complaints, number of sexual partners, last sexual intercourse, condom use, injection of TPHA/TPPA, and referral to VCT. To determine factors that most affect transmission of IMS, this research uses analysis Multivariate Adaptive Regression Spline (MARS) method. MARS has several advantages including flexible for high dimensional data modeling. Having obtained the MARS models, the classification will be done to determine the classification accuracy of the MARS models. The results show that the MARS approach obtains four variables that affect (contribute) the response variable, which are marital status of PSK, complaints of IMS, total couple sex (last week) and use of condoms (last week). The classification accuracy of MARS method is 64,90%.