Prediction of Insufficient Accuracy for Mushroom Classification whether Poisonous or Eatable Food using Random Forest Training by comparing Logistic Regression to Improve Accuracy

Abstract

Aim: Mushrooms have a spot with development type and they contain principal supplements, for instance, proteins, nutrients,
cell reinforcements, antioxidants and amino acids. There are a huge load of benefits of mushrooms. A wide range of mushrooms
are not eatable. So before eating up mushrooms, it should be checked for consumable mushrooms. Careful affirmation and fitting
distinctive confirmation of species are the super protected way to deal with ensuring not eatable mushrooms, and safeguard
against expected accidents of consuming poisonous one. Materials and Methods:The survey used 51 models with two get togethers of computations with the G-power worth of 80% and the Mushroom were accumulated from various web sources with late audit disclosures and edge 0.05%, sureness length 90% mean and standard deviation. To expect the Mushroom precision rate for at present the Logistic Regression computation has found 90.57% of accuracy, therefore this study needs to find better accuracy for accuracy decrease prediction with the novel Random Forest Training Machine Learning estimation. Result: This investigation found 94.64% of accuracy for poisonous distinguishing proof using the Random Forest Training estimation with a statistically significant difference between the two groups (p=0.047; p<0.05) with 95% sureness stretch. Conclusion:This investigation found 92.82% of accuracy for poisonous recognizable proof using the Random Forest Training estimation with a basic worth of two followed tests is 0.001(p<0.05) with 95% conviction stretch.