Machine Learning-Based Irrigation Water Demand Forecasting
- Author
- Chinhema, Jalin. P
- Title
- Machine Learning-Based Irrigation Water Demand Forecasting
- Abstract
-
The efficient management of water resources in agriculture is critical for sustainable farming practices. This research project focuses on the development and evaluation of a machine learning-based model for forecasting irrigation water demand. The study aims to analyze different deep learning techniques, design and implement a deep learning model, and evaluate the effectiveness of machine learning techniques in irrigation water demand prediction. To achieve these objectives, the researcher utilized the Random Forest regression algorithm to develop a model capable of predicting seasonal water demand for a farm. Black-box, white-box, and performance tests were conducted, including the use of the confusion matrix, to assess the system's performance. The results showed satisfactory performance, with an accuracy of 98.5% and a classification error rate of 0.025%. The model achieved an overall precision of 100% and a recall of 95%, along with an F1 score of 97.5% and a true negative rate of 98%. Validation accuracy stood at 95%, and a mean percentage error of -0.044% was achieved. Comparative analysis with previous research studies indicated significant improvements, surpassing the results obtained by other researchers using techniques such as Multiple Linear Regression (MLR), Principal Component Regression (PCR), and other models. The LSTM deep learning architecture employed in this research achieved an average improvement of 7%-9% compared to previous studies. This research contributes to enhancing the accuracy and reliability of irrigation water demand forecasting. The developed model offers valuable insights for farmers and policymakers, aiding in better water resource management and sustainable agricultural practices. The findings highlight the potential of machine learning in optimizing irrigation practices, leading to improved resource allocation and reduced water wastage in the agricultural sector
- Date
- 2023
- Publisher
- BUSE
- Keywords
- Machine Learning-Based Irrigation Water Demand Forecasting
- Supervisor
- Mr Musariwa
- Item sets
- Department of Computer Science
Part of Machine Learning-Based Irrigation Water Demand Forecasting