Anticipating the long-term effects of daily weather conditions and socio-economic factors on FNN through metropolitan yard waste

Abstract

The ability to accurately forecast the ability of a system to function is necessary for efficient and productive management of solid waste. The correctness of wastage forecast proposals, which have been the subject of extensive research, is tested. In contrast to wastage forecast proposals for varied waste products, factors for yard wastage are time-sensitive and the repercussions of delay must be considered. In this work, we focus on lag periods concerning factors that aim to forecast the creation of metropolitan yard wastage using the Deep Learning (DL) technique. Through correlation evaluation, socioeconomic and meteorological factors are evaluated on a daily average and the key factors are used to develop yard wastage proposals. Following that, these proposals employ Feed Forward Neural Networks (FNN), that variety of day measurements are represented by time-stamped variables. This facilitates to achieve the decrease of the daily yard wastage production's inaccuracy. We developed an FNN model that can forecast the influence of these factors on yard wastage generation and management methods using a wastage dataset that includes meteorological variables, socioeconomic indicators and distributions of yard wastage from a metropolis region. At the training and testing stages of yard wastage models, we compare the results using MSE, MAPE and R-Value at each stage's perfect number of layers. The effect of delay on the accurateness of daily yard wastage forecast proposals was highlighted by the MAPE of 25.35% that was attained during testing and the 23.2% reduction in the MSE displayed by yard wastage proposals during training.