The Novel Method About Correlation Calculation Based on An Inequation

Among the many methods of analysing correlation coefficient, the two classical methods are the Spearman correlation coefficient and Pearson correlation coefficient. For the Spearman method, the time complexity is larger because of the sorting data. For the Pearson method, there are also some limited conditions, such as the data are needed to linearly related and continuous, normally distributed or near-normally distributed with a single peak and independent. To avoid the disadvantages about the above correlation coefficients, we propose a novel method about correlation calculation based on an inequation, which derives from inventory model theory and risk pooling theory. Firstly, for the whole dataset, we calculate the sample mean and standard deviation of sub elements and whole datasets respectively. Then, using the inventory model theory, we calculate the total standard deviation of the whole dataset and the sum of standard deviation about sub elements of whole dataset, and obtain the novel inequality. Then, judge this inequality: when “\(=\)” is true, conclude that there is no correlation data in the entire dataset \({\rm{X}}\); when “\(>\)” is true, it is judged that the entire dataset X has relevant data, furthermore, the bigger the difference of left value subtracts right value, the stronger the correlation. The proposed algorithm owns the lower time complexity and much wider range of uses. Finally, the GDP experiment and Gini index experiment demonstrate the effectiveness of the novel method.