Extreme learning machine model for assessment of stream health using the habitat evaluation index

Abstract Extreme Learning Machine (ELM) approach is used to predict stream health with Qualitative Habitat Evaluation Index (QHEI), and watershed metrics. A dataset of 112 sites in Ontario, Canada their Hilsenhoff Biotic (HBI) richness values two ELM models development. Each model 70 30% the for training testing respectively. The show a great fit Root Mean Square Error (RMSE)=0.12 0.33 HBI test models. Then, features elimination based on coefficients coefficient variation slight increase models' RMSE reach 0.09 correspondingly. Accordingly, this high predictabilities research provide better insights which factors influence or Richness that has architecture than other machine learning ANN learn complex non-linear relationships. Also, sensitivity analysis expressed channel slope as most affecting stream-health parameter health.