Advantages of Synthetic Noise and Machine Learning for Analyzing Radioecological Data Sets

Data set I: Fungi at Chernobyl The number of fungal taxa isolated from soil at various locations near the Chernobyl nuclear power plant during the first five years after the accident (the outcome, called TotalTaxa) had a moderate negative correlation with the severity of radioactive contamination (AvLogRad) (Fig 1). The other numerical predictor variables (Depth, pH and Year) had small positive correlations with the outcome (Fig 1). The Mantel test revealed no auto-correlation of the number of taxa with depth in soil (p-value = 0.98). Poisson regression showed that AvLogRad had a statistically significant negative effect (p-value = 0.006), whereas all other predictors did not achieve statistical significance (Table 2). MMI assigned the highest importance to AvLogRad, followed by Year and Depth (Table 2). Because there was evidence for collinearity…


Link to Full Article: Advantages of Synthetic Noise and Machine Learning for Analyzing Radioecological Data Sets