Rankings Correlation Study : Brand Search Volume vs. Brand Sales Volume
Chenyang Wang, Chih‐Yu Hsu · 2020
Hadoop MapReduce is a commonly used data processing engine for big data analysis. An iterative MapReduce parallel computing workflow was designed to analyze the logs of search engines in fast-moving consumer goods (FMCG) e-commerce websites. Based on the search results of the brand facet in each search, the workflow mines the potential correlation between each keyword search and brand awareness. In the workflow, the contribution value of each search to the search popularity of each brand in the brand facet search result is calculated to obtain a list of contribution value. Then, the contribution value list is integrated to calculate the search volume rank-order of each brand. If higher rank-order of brand search volume tend to have higher rank-order of the brand sales volume, this fact can help companies design their sales strategies. Therefore, it is important to test the relationship between the rank-order of the two types of volumes. The Spearman's correlation coefficient measures the rank-order relationship between brand search volume and brand sales volume to show the correlation is significant at the 0.01 level (two-tailed). There is a strong and positive rank-order correlation between brand search volume and brand sales volume, which was statistically significant ($\rho$=.770, p=.009).