Patterns of waste collection: A time series model for market waste forecasting in the Kumasi Metropolis, Ghana

Abstract

Forecasting waste quantities is vital for effective and sustainable waste management although waste data, particularly for market waste (MW), is inadequate in many developing countries. Accurate prediction of waste aids in waste budgetary planning with optimized waste management techniques. The research found the total MW collected during the wet and dry seasons from six major markets in Kumasi, the second-most populated city in Ghana. The markets are Central, Bantama, Ayigya, Suame, Tafo and Moro. The trends over a 132-monthly period were also studied. The study utilizes Panel-Corrected Standard Errors (PSCEs) to determine the significant relationship between waste quantities and four explanatory variables: market population estimate; MPE, the distance between communal container and landfill; DIST, waste container lifting frequency; CLF and a number of communal waste containers; NOC. Time Series ARIMA model forecasts waste quantities collected in the Kumasi markets. Findings indicated an unanticipated gradual fall in annual MW quantities, from 2014 to 2018, with increasing years and attributed this to raised awareness on waste reuse and recovery practices among traders and communal dumpsite attendants, indiscriminate waste dumping by proliferated auto-rickshaw operators, and improper record keeping at the landfill. A significant relationship (R² = 59 %) between waste quantities and MPE, CLF, and DIST variables is noted from the PCSEs model. The article further demonstrated that ARIMA (4,1,3) for HDM and ARIMA (3,1,3) for LDM were robust forecasting models for MW quantities collected in Kumasi. The fitted models forecasted MW quantities for 132 months (2011–2029) to be approximately 335 thousand tonnes.