An improved credibility method based on matrix weight

While in many fields the credibility method has obtained great successful applications, but there are still many shortcomings. This paper presents a credibility method based on matrix weights analysis, an improvement focusing on credibility method based on Euclidean distance analysis within the normal distribution. It not only solves the contradictory conclusion due to the too close numerical value between the prior probability and posterior probability, but also reduces the negative effects from non-important evidence, which better reflects the positive effect of important evidence, eliminating the defects of traditional credibility methods in these two aspects.