Skip to main content

Table 5 Examples of application of dimension reduction methods towards SDGs using EO data

From: Monitoring sustainable development by means of earth observation data and machine learning: a review

SDGs Field Main finding References
SGD 2 (Zero Hunger) Agriculture Partial Least Square Regression was applied with success, as a FS method, on crop yield estimation [115]
The FS results demonstrated that the proposed Maximum Separability and Minimum Dependency method was more accurate than filter methods [116]
SGD 6 (Clean Water and Sanitation) Water resources The proposed approach proved to be effective and accurate to assess water resources at catchment scale [117]
Water Sources Stepwise Discriminant Analysis and PCA improved the accuracy of water source recognition [118]
SDG 7 (Affordable and Clean Energy) Electricity The proposed method improved the forecasting of electricity price and it was more accurate than the Independent Electricity System Operator prediction [119]
SGD 9 (Industry, Innovation and Infrastructure) Structural Reliability The Bivariate Dimension Reduction Method proved to be effective for structural reliability analysis [120]
SGD 11 (Sustainable Cities and Communities) Land Cover The results demonstrated that FS improves the classification accuracy of land cover classification [121]
The proposed method demonstrates better results compared to other methods for land cover classification in almost all tests [122]
Dimensionality Reduction was considered a key step in the land cover classification process [123]
The experiments shown that the impervious surface extraction accuracy of Classification and Regression Tree was higher than Seperability and Thresholds algorithm [124]
Land use FS with Classification Optimisation Score metric reduces the processing time and produces higher classification accuracy for land use and land cover classification using VHR data [125]
SDG 13 (Climate Action) Pollution The new Dimension Reduction method demonstrated to be a powerful approach to optimise the knowledge that emerges from atmospheric observations of N2O [126]
SGD 15 (Life on Land) Forest Proposed a FS SVM-Recursive Feature Elimination method to explore the relationship between the biomass and parameters derived from Landsat-8 imagery. The results demonstrated that this method was able to accurately estimate the aboveground biomass. [127]
Terrestrial ecosystem FS methods allow the extraction of valuable information to create accurate maps of areas infested by invasive plant species [128]