Tehrani, Rouzbeh Afsarmanesh; Faheem, Ahmed; Fardmanesh, Mohsen; Ranjbar, Sibia; Suri, Rominder P. S. (Temple University. Libraries, 2016)
      Waste disposal has always been one of the challenging aspects of human life mostly in populated areas. In every urban region, various factors can impact both amount and composition of the generated waste, and these factors might depend on a series of parameters. Therefore, developing a predictive model for waste generation has always been challenging. We believe that one main problem that city planners and policymakers face is a lack of an accurate yet easy-to-use predictive model for the waste production of a given municipality. It would be vital for them, especially during business downturns, to access a reliable predictive model that can be employed in planning resources and allocating budget. However, most developed models are complicated and extensive. The objective of this research is to study the trend of solid waste generation in Philadelphia with respect to business cycle indicators, population growth, current policies and environmental awareness, and to develop a satisfactory predictive model for waste generation. Three predictive models were developed using time series analysis, stationary and nonstationary multiple linear regressions. The nonstationary OLS model was just used for comparison purposes and does not have any modeling value. Among the other two developed predictive models, the multiple linear regression model with stationary variables yielded the most accurate predictions for both total and municipal solid waste generation of Philadelphia. Despite its unsatisfactory statistics (R-square, p-value, and F-value), stationary OLS model could predict Philadelphia’s waste generation with a low level of approximately 9% error. Although time series modeling demonstrated a less successful prediction comparing to the stationary OLS model (25% error for total solid waste, and 10.7% error for municipal waste predictions), it would be a more reliable method based on its model statistics. The common variable used in all three developed models which made our modeling different from the Streets Department’s estimations was unemployment rate. Including an economic factor such as unemployment rate in modeling the waste generation could be helpful especially during economic downturns, in which economic factors can dominate the effects of population growth on waste generation. A prediction of waste generation may not only help waste management sector in landfill and waste-to-energy facilities planning but it also provides the basis for a good estimation of its future environmental impacts. In future, we are hoping to predict related environmental trends such as greenhouse gas emissions using our predictive model.