Water Quality Can Be Forecasted Using Machine Learning
A study published in the International Journal of Sustainable Agricultural Management and Informatics has shown the potential of machine learning to forecast the water quality index, which could have significant implications for water management in both drinking water and agricultural applications.
Water pollution, a concern of recent times
The decline in water quality has become a matter of concern in recent times, with a growing focus on its effects on human health and agricultural output. Currently, the improper discharge of untreated sewage causing pollution in rivers and coastal waters is a significant environmental concern. In contrast, the matter of water security in agriculture remains a topic of importance.
Water quality is evaluated based on several factors: acidity and alkalinity, pH level, turbidity, dissolved oxygen, nitrate concentration, temperature, and fecal bacteria. To manage and regulate pollution is essential to create efficient techniques for predicting water quality.
Machine learning in favor of water quality
At the Higher Institute for Applied Sciences and Technology (HIAST) in Damascus, Syria, Ahmad Debow, Samaah Shweikani, and Kadan Aljoumaa have designed 4-stacked LSTM models for the anticipation of WQI.
A 4-stacked LSTM (Long Short-Term Memory) refers to a recurrent neural network capable of identifying long-term data patterns that change over time. After analyzing the data, such models can make projections regarding future alterations in that data. By placing four LSTM layers sequentially, the model can more effectively detect subtle patterns in the data.
The team employed various algorithms to arrange the data and select relevant features for analysis, such as K-NN (K nearest neighbors) and annual mean. K-NN is a widely recognized algorithm in machine learning used for classification and regression purposes.
It is a non-parametric algorithm that does not presume any specific characteristics about the underlying data distribution. The fundamental concept behind K-NN is to categorize new data points by assessing their similarity with the closest neighbors in the training dataset.
The team’s accomplishment in reproducing established data through these models is encouraging for real-world projections and has the potential to contribute to water management initiatives significantly.
By utilizing the forecasts made by these models, it should be possible to implement more preemptive measures to reduce pollution in the water supply for human consumption and agricultural purposes.
By utilizing the forecasts made by these models, it should be possible to implement more preemptive measures to reduce pollution in the water supply for human consumption and agricultural purposes.
Read the original article on PHYS
Read More: Research Sheds New Light on the Benefits of Water Fluoridation to Children
