REGIONAL PATTERN OF HIV CASES IN PAKISTAN
Keywords: Forecast, HIV cases, Pakistan, Time series models
AbstractObjective: To forecast the HIV cases by location and geographical area of res- idence in Pakistan. Methodology: The record of monthly number of HIV cases from 18 HIV treat- ment centres for the period of January 2013 to July 2014 taken from Pakistan Bureau of Statistics (PBS). Descriptive frame work along with time series smooth- ing models (single parameter exponential smoothing, double parameter expo- nential smoothing, Holt’s and Winter’s multiplicative model) were applied. Results: On the average Punjab (3234 ±556) and Sindh (2606 ±345) had higher monthly HIV cases, while the HIV cases in Islamabad were more consistent due to least coefficient of variation (7%). Among the competing time series models Winter’s (overall, Islamabad, Punjab, Sindh, Balochistan) and Holt’s (Khyber Pa- khtunkhwa) models were found to be superior in forecasting the HIV cases due to least forecasting error. The increasing trend in HIV cases is expected for all over Pakistan (2.2%), Islamabad (1.3%), Punjab (3.4%), Sindh (1.9%), Balochistan (4%) and Khyber Pakhtunkhwa (1.3%). Conclusion: Increasing trend is expected for HIV cases in Pakistan.