A study of Fuzzy Probability Framework for validating Mineral Nutrients in Compost Sample

Abstract

This paper illuminates on the employment of a machine learning-based fuzzy probability
approach to handling a huge quantity of imprecise, uncertain, redundant and inconsistent data, decides its
quality and standardization of a particular data point in the Municipal Solid Waste (MSW) compost sample
dataset. Fuzzy probability finds uncertainties and filter out certain data points by characterizing the
uncertainty with probability. While, machine learning typically looks for pattern recognition that predicts
likely to happen events, in the sense the probabilistic classification can be easily done to bring the concept
from generalization to specification. Introducing machine learning in decision making to produce desired
output assists in predicting the values more accurately. The paper proposed a theoretical study on the design
of Certainty-Validity Framework (CVF) to merges machine learning with fuzzy probability theory as well as
its implementation on compost samples. The paper also demonstrates the use of a neural network classifier
to assign membership to each data point in a fuzzy set to find certainty. Furthermore, the result shows the
design of inference rules to find validity Vi for the validating data point by mapping membership and
probability. Summation of all Vi’s form score value Si, thus standardizing the mineral nutrient in compost
sample data. The conclusion states that the proposed theoretical framework can be used in various other
applications where critical and sensitive dataset is used