الملخص الإنجليزي
Abstract:
Excess domestic waste generation during religious festivals has
adverse environmental impacts. The holy month of Ramadan provides
opportunity to study relationship between food consumption pattern and
waste generation. Therefore, the objective of this research is to
investigate the relationship between Ramadan fasting and the generation
of domestic solid waste in Bahrain for the period between 2019-2022.
Furthermore, this study develops accurate time-series models using deep
machine learning approaches to predict domestic waste generation for
pre-, during, and post-Ramadan periods. The research findings revealed
that during the month of Ramadan, the Bahraini's generates the highest
amount of waste, with an average of 1,570 tons per day, compared to
1,450 tons per day during the rest of the year. The study conducted a
comparative analysis of three machine learning models: Long Short-Term
Memory (LSTM) Recurrent Neural Networks, Bidirectional LSTM
(BILSTM) Recurrent Neural Networks, and Gated Recurrent Unit
(GRU), along with three statistical models: Autoregressive Integrated
Moving Average (ARIMA), Seasonal Autoregressive Integrated Moving
Average (SARIMA), and Prophet. The prediction of BILSTM model
outperformed the others with MAE value of 28.120, RMSE value of
36.879, MAPE value of 0.020, and R2 value of 0.78.
