Determining Statistical Pattern on the Drug-Related Killing in Philippines Using ARIMA and Poisson Techniques

A univariate time series technique was conducted to determine statistical pattern on killings of drug suspects in Philippines from May 19 to July 7, 2016. The technique reveal a moving-average of order 2, MA(2) with a positive coefficient suggestive that value of outcome variable tend to increase, on the average, than the recent value of . This means that drug-related killings will tend to be higher than the most current rate; and killings is seen to increase as weekend comes. Poisson regression indicated an average of 13 deaths on a Sunday; only 2 on average on a Monday; odds of survival increases as well as weekend comes. Finally, the forecast model and the simulation are limited by the data used. Structure of the univariate series may change as additional data are added; this is also true for the forecasted average occurrence.