A comparison of unsupervised and supervised machine learning algorithms to predict water pollutions

Zamri, Nurnadiah; Pairan, Mohammad Ammar; Azman, Wan Nur Amira Wan; Abas, Siti Sabariah; Abdullah, Lazim; Naim, Syibrah; Tarmudi, Zamali; Gao, Miaomiao

doi:10.1016/j.procs.2022.08.021

Cited by 16 publications

(6 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The research used the k-means clustering algorithm, a well-established method in data analysis for identifying patterns in datasets, notable for its guaranteed convergence (Piegari et al, 2023). According to Zamri et al, (2022), the elbow and silhouette approaches were used to uncover clusters in the dataset that might reveal concealed subsurface units. The datasets underwent principal component analysis (PCA) to reduce the dimensions while retaining key traits and reducing complexity.…”

Section: K-mean Clusteringmentioning

confidence: 99%

See 1 more Smart Citation

Joint K-means clustering and statistical analytical modeling of P- wave velocity and resistivity datasets for subsurface lithologic differentiation

Dick,

Bery,

Bala

et al. 2024

Preprint

View full text Add to dashboard Cite

Given the hazards linked to unstable ground conditions, it is vital to grasp the soil-rock characteristics essential for foundation construction and groundwater development. However, the inherent challenges in geophysics, such as the non-uniqueness of the inverse problem and incomplete subsurface knowledge, hinder the direct interpretation of geophysical data in terms of geological units. Traditional soil exploration methods or relying solely on one geophysical survey method often yield inaccurate results due to limitations in mapping subsurface complexities and heterogeneities. This study addresses these challenges by applying K-means cluster analysis to a univariate geophysical parameter set spanning an 800 m section in the geothermally active Kabota-Tawau area of Sabah, Malaysia. Leveraging unsupervised machine learning techniques like principal component analysis, involving Silhouette and elbow methods, the research determines the optimal number of clusters (k) and validates their accuracy. The analysis identifies four distinct lithologic units, serving as proxies for soil/rock properties in the study area. With an R-squared value nearing 1 and an average Silhouette score of 0.67 for \(k=4\), the results indicate a high level of satisfaction in cluster separation, supported by a percentage sum of square error exceeding 88%. This approach enhances our ability to accurately identify lithologic units critical for improving the reliability of foundation construction and groundwater development efforts.

show abstract

Section: K-mean Clusteringmentioning

confidence: 99%

“…Using models of P-wave velocity (Vp) and resistivity with ML in the same place for subsurface exploration makes it possible to nd a clear link between physical properties at certain sites (Piegari et al, 2023;Zamri et al, 2022). This approach facilitates the classi cation of the surveyed region based on its distinctive lithology.…”

Section: Introductionmentioning

confidence: 99%

Joint K-means clustering and statistical analytical modeling of P- wave velocity and resistivity datasets for subsurface lithologic differentiation

Dick,

Bery,

Bala

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Its purpose is to ensure uniformity in scale and value distribution across all variables, thereby promoting equitable contributions to model development and diminishing potential bias stemming from variable dominance. In this context, the data underwent a transformation to attain a mean of zero and a stan-dard deviation of 1, employing the StandardScaler method available in the Scikit-learn library [25,49,50].…”

Section: Pre-processing Of Data and Statistical Analysismentioning

confidence: 99%

“…Nonetheless, machine learning algorithms rely on a single estimation model and may tend to overfit when confronted with limited training data, as observed with KNN, SVM, and RF [8]. To alleviate such concerns, pre-processing input data through normalization, utilizing extensive databases [25], and applying data partitioning methods like K-fold [26] can effectively mitigate the risks of overfitting and underfitting during the training process, thereby bolstering the models' capacity for generalization. Integrating machine learning with data from UAVs and satellites stands as a promising strategy for monitoring plant height dynamics at a large experimental scale.…”

Section: Introductionmentioning

confidence: 99%

Integrating Satellite and UAV Technologies for Maize Plant Height Estimation Using Advanced Machine Learning

Ferraz,

Barboza,

Arantes

et al. 2024

AgriEngineering

View full text Add to dashboard Cite

The integration of aerial monitoring, utilizing both unmanned aerial vehicles (UAVs) and satellites, alongside sophisticated machine learning algorithms, has witnessed a burgeoning prevalence within contemporary agricultural frameworks. This study endeavors to systematically explore the inherent potential encapsulated in high-resolution satellite imagery, concomitantly accompanied by an RGB camera seamlessly integrated into an UAV. The overarching objective is to elucidate the viability of this technological amalgamation for accurate maize plant height estimation, facilitated by the application of advanced machine learning algorithms. The research involves the computation of key vegetation indices—NDVI, NDRE, and GNDVI—extracted from PlanetScope satellite images. Concurrently, UAV-based plant height estimation is executed using digital elevation models (DEMs). Data acquisition encompasses images captured on days 20, 29, 37, 44, 50, 61, and 71 post-sowing. The study yields compelling results: (1) Maize plant height, derived from DEMs, demonstrates a robust correlation with manual field measurements (r = 0.96) and establishes noteworthy associations with NDVI (r = 0.80), NDRE (r = 0.78), and GNDVI (r = 0.81). (2) The random forest (RF) model emerges as the frontrunner, displaying the most pronounced correlations between observed and estimated height values (r = 0.99). Additionally, the RF model’s superiority extends to performance metrics when fueled by input parameters, NDVI, NDRE, and GNDVI. This research underscores the transformative potential of combining satellite imagery, UAV technology, and machine learning for precision agriculture and maize plant height estimation.

show abstract

“…The main reason was that the water quality in the flood season was uncertain due to the influence of Water pollution prediction has always been a hot research topic for scholars. Zamri et al compared the performance of 10 machine learning algorithms in pollution source classification, and found that RF algorithm has high performance in pollution source classification prediction (with 98.78% accuracy) [51]. The main reason was that RF can integrate multiple decision tree models to provide more stable prediction.…”

Section: Plos Onementioning

confidence: 99%

Integrated prediction of water pollution and risk assessment of water system connectivity based on dynamic model average and model selection criteria

Ruan,

Cui,

Meng

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

In recent years, with the rapid development of economy and society, river water environmental pollution incidents occur frequently, which seriously threaten the ecological health of the river and the safety of water supply. Water pollution prediction is an important basis for understanding development trends of the aquatic environment, preventing water pollution incidents and improving river water quality. However, due to the large uncertainty of hydrological, meteorological and water environment systems, it is challenging to accurately predict water environment quality using single model. In order to improve the accuracy and stability of water pollution prediction, this study proposed an integrated learning criterion that integrated dynamic model average and model selection (DMA-MS) and used this criterion to construct the integrated learning model for water pollution prediction. Finally, based on the prediction results of the integrated learning model, the connectivity risk of the connectivity project was evaluated. The results demonstrate that the integrated model based on the DMA-MS criterion effectively integrated the characteristics of a single model and could provide more accurate and stable predictions. The mean absolute percentage error (MAPE) of the integrated model was only 11.1%, which was 24.5%–45% lower than that of the single model. In addition, this study indicates that the nearest station was the most important factor affecting the performance of the prediction station, and managers should pay increased attention to the water environment of the control section that is close to their area. The results of the connectivity risk assessment indicate that although the water environment risks were not obvious, the connectivity project may still bring some risks to the crossed water system, especially in the non-flood season.

show abstract

A comparison of unsupervised and supervised machine learning algorithms to predict water pollutions

Cited by 16 publications

References 9 publications

Joint K-means clustering and statistical analytical modeling of P- wave velocity and resistivity datasets for subsurface lithologic differentiation

Joint K-means clustering and statistical analytical modeling of P- wave velocity and resistivity datasets for subsurface lithologic differentiation

Integrating Satellite and UAV Technologies for Maize Plant Height Estimation Using Advanced Machine Learning

Integrated prediction of water pollution and risk assessment of water system connectivity based on dynamic model average and model selection criteria

Contact Info

Product

Resources

About