Agenda 12/02/2004 - Laserfiche WebLink

Home Browse Search

Page Two November 30, 2004 occurred in the community since the year 2000 any data prior to 2000 should not be used to forecast the future. If you look at the City's jail population from 2001 - 2004 you will see very little change in the inmate population. This makes sense when you take information provided by the City's Police Chief showing crime in the City has been decreasing and continues to do so coupled with a declining and aging population. With a rebounding economy this trend should continue as more and more of our citizens return to full employment even if the population increases and becomes younger. The City of Martinsville saw the problem of prison overcrowding almost 10 years ago. In a joint meeting between the City Council and the Henry County Board of Supervisors on March 29, 1995 the topic was discussed. The result was the Board of Supervisors declined to join any regional jail effort and the City decided to pursue a City - only option to solve its over crowding problem. In the process of preparing our application for expansion to the Virginia Department of Corrections (VDOC) the City Manager sent a letter dated July 25, 1996 asking the Counties of Carroll, Floyd, Franklin, Henry, Patrick, and Pittsylvania and the City of Danville and the Town of Rocky Mount if they would like to participate in the project, all declined. The city then constructed what is referred to as Building #2. Consultants have an inherent and at times financial interest in over estimating the number of inmates thus the size of the jail project. Often they are paid based on the size of the project and the larger the project the higher their fees. Also, no consultant wants to face the wrath of any community if they recommend and the community builds a facility that rapidly becomes to small causing the community to build another facility in the near future. This has been the case in both our Jail needs study and the study involved in the building of the W. W. Moore detention center. In our Jail study the consultants lowest estimate showed the City of Martinsville would have 167 inmates by 2003. This is 33 or 25% more than our actual average for the year. Using the Exponential Smoothing forecasting method is not sound. As stated in the textbook "Basic Statistics in Business and Economics' by George Summers, William S. Peters, and Charles Armstrong, Exponential Smoothing "is often used for very short-term forecasting, that is, forecasting an activity level for the next week, next month, or possibly the next quarter." To use it to forecast 14 years into the future is not recommended. I spoke to Mr. Powell and asked why he used either the Exponential Smoothing or Box- Jenkins models for all the communities when his own rePOrt clearly showed that linear regression was a better predictor for some of the communities. He stated the Virginia Department of Corrections (VDOC) like to use those two models as compared to Linear Regression. He admitted Linear Regression would often produce lower forecasts than the other two models. I would recommend using linear regression to forecast future inmate population. As can be seen in the first attachment this model