Combining Multiple Machine Learning Algorithms to Predict Taxa Under Reference Conditions for Streams Bioassessment

Abstract

In the present study, we tested the potential of combining three machine learning techniques in a bioassessment tool to more accurately predict the pool of expected taxa at a site. This tool, the Hydra, uses the best performing technique from Support Vector Machines (SVM), Multi‐layer Perceptron and K‐Nearest Neighbour (KNN), to predict the taxa expected at a stream site, and further evaluates the quality of a site, though a classification system based on observed/expected values, similar to that used in River Invertebrate Prediction and Classification System (RIVPACS) models. To test the procedure, we used a dataset composed of 137 training sites, 15 validation sites and 174 test sites (potentially disturbed) from Portuguese streams. The combined use of three machine learning techniques was more effective in the prediction of invertebrate taxa at a site than their individual use. The three methods were always tested for all invertebrate taxa, but from the three techniques tested, SVM and KNN were most often the best performing techniques (the most accurate among the three for a higher number of taxa) in the prediction of invertebrate taxa with the present dataset. The combination of all algorithms implemented in Hydra resulted in good models for stream bioassessment (e.g. SD OE50 < 0.2, regression of O vs E: R2 > 0.6, Spearman correlations with global degradation >0.7). We also found no advantage in removing rare taxa from the training dataset, and 50% accuracy is the most adequate accuracy level for calculation of OE ratios through Hydra. Future work should consist of comparing the performance of this technique with others, such as the RIVPACS models, using the same datasets. Considering the flexibility of this technique, self‐adjustment and easy implementation through a website (aquaweb.uc.pt), we expect it to be also useful in the prediction of other aquatic elements such as fishes and algae. Copyright © 2013 John Wiley & Sons, Ltd.

Advances in Watershed Science and Assessment

“…Major plant species within a complex, heterogeneous wetland have been classified using multi-temporal high-resolution QuickBird satellite images, field reflectance spectra, and Lidar height information [109]. Using Lidar, hyperspectral and radar imagery, and narrow-band vegetation indices, researchers can discriminate between some wetland species and also estimate biochemical and biophysical parameters of wetland vegetation, such as water content, biomass, and leaf area index [69,[109][110][111][112].…”

“…Machine learning techniques, including artificial neural networks (ANNs) and ensemble prediction trees, where models are iteratively trained at prediction to minimize error, have received much recent attention for predictive modeling in ecology [107][108][109][110]. Though the method is not yet widely used, support vector machines have performed favorably compared with other machine learning techniques for predicting the occurrence of macroinvertebrate taxa [111,112].…”

Rehabilitation scenarios for reservoirs: Predicting their effect on invertebrate communities through machine learning

“…Accordingly, simulation tools are key for assessing successful recovery because they can incorporate multiple information sources in degradation indices and test different combinations of rehabilitation scenarios (Brudvig, 2017; Hermoso et al, 2012). Predictive models based on machine learning techniques have been developed and tested for bioassessment of rivers and reservoirs (e.g., Chen and Liu, 2014; Feio et al, 2020; Feio, Viana‐Ferreira, & Costa, 2014a, 2014b; Gabriels, Goethals, Dedecker, Lek, & De Pauw, 2007; Linke, Norris, Faith, & Stockwell, 2005; Park, Cho, Park, Cha, & Kim, 2015; Sarrazin‐Delay, Somers, & Bailey, 2014), and have been shown to be promising tools in the context of river rehabilitation. These approaches have the ability to model and predict species distribution in dimensional space with advantages over classical predictive modelling techniques of: not requiring a priori reference sites that can be viewed as artificial; capturing nonlinear relationships; and being less influenced by outliers (Gevrey et al, 2004; Rose, Kennard, Moffatt, Sheldon, & Butler, 2016).…”

“…The HYDRA machine learning tool (Feio et al, 2014a, 2014b) is a modelling tool that allows the prediction of communities through the simultaneous use of three different techniques: support vector machines (SVM;), multilayer perceptron (MLP) and K‐nearest neighbour analysis (KNN). This tool has already been widely tested with large datasets from Europe, North America and Oceania rivers and has displayed a good precision and accuracy in identifying alterations in aquatic communities (Feio et al, 2014a, 2014b, 2020).…”

Dangerous relationships: biases in freshwater bioassessment based on observed to expected ratios

“…, Feio et al. , b, Rose et al. , b), or the approach applied apparently independently (e.g., Hargett et al.…”