1. Application of artificial neural networks for predictive model of municipal solid waste collection in tourist cities.
- Author
-
Na Ayutthaya, T. Kridakorn, Jakrawatana, N., Rinchumphu, D., and Owatsakul, V.
- Subjects
ARTIFICIAL neural networks ,SOLID waste management ,REFUSE collection ,DATA mining ,STANDARD deviations - Abstract
BACKGROUND AND OBJECTIVES: Tourism is a critical component of the economic framework in Thailand and numerous other countries worldwide, acting as a significant revenue generator. In the year 2022, the tourism industry made a significant contribution to Thailand's Gross Domestic Product, representing 22.92 percent of the total. However, the emphasis on urban development and management in order to boost visitor numbers can worsen urban metabolism, leading to an escalation in resource. The urban administration needs to know how much waste must be managed to plan for effective environmental and public health management. The focus of this study is on the construction of an artificial neural network forecasting model that takes into account socioeconomic and demographic variables to anticipate the management of municipal solid waste in a city known for tourism. METHODS: Models were generated by synthesizing and integrating municipal solid waste collection quantities with 17 inputs of socioeconomic, demographic, event occurrences, and tourism-specific metrics variables from 2015-2021 in Chiang Mai municipality. Deep learning techniques were used to create the models. Socioeconomic characteristics were derived using nosecount data at the provincial and municipal levels. Data preprocessing involved the implementation of knowledge discovery in database strategy to ensure the creation of datasets with sufficient numbers and quality for modeling. This issue involved calculating the correlation coefficient between 17 inputs and the quantities of municipal solid waste collected. RapidMiner® computer software was used to construct a model incorporating frameworks using artificial neural network techniques. To ensure robustness and prevent overfitting, the dataset was divided into training and validation sets. The model was trained using backpropagation methods, and the evaluation of the model's performance was based on the correlation between the observed and predicted values of the mean municipal solid waste collection rate. FINDINGS: The waste prediction model achieved optimal performance by incorporating eight input variables across two hidden layers, one consisting of ten nodes and the other of five nodes. Across eight trials, this arrangement produced the lowest correlation coefficient (0.67), mean absolute error (320.779 +/- 22.080), and root mean square error (16.5). On the other hand, the chosen model used 17 input variables split across two hidden layers, each with 8 or 4 nodes. The model yielded a correlation coefficient of 0.69, a mean absolute error of 461.953 +/- 706.680, and a root mean square error of 21.9. The current daily amount of municipal solid waste collected is 340 metric tonnes, while the projection model anticipates an increase to 348 metric tonnes per day by 2023, with a margin of error of 2 percent. The model further predicts a daily garbage collection of 361 metric tons through 2030. CONCLUSION: Future waste management strategies may be planned, and various environmental impacts in tourism cities can be analyzed using the forecasting process and framework for the municipal solid waste collection rate described in this research. These characteristics are harnessed by the model to gain a thorough insight into waste dynamics in metropolitan regions with high tourist activity, ultimately facilitating the adoption of more sustainable urban planning and management approaches. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF