Analysis of Sales data for the Semiconductor using Data mining

Abstract

Increased demand for products which are Smartphone, tabletPC and other mobile device using Mobile DRAM over the world makes sales increase of Mobile DRAM. This paper suggests valid statistical methods for application of sales data and analyzes the relation and trend among the type of Mobile DRAM, density and sales area. In addition, we could get another new idea via the result. For analysis, Clustering, logistic regression with lasso, decision tree and Partial Correlation Estimation method are introduced. glasso (graphic lasso) that is algorithm to estimate a sparse inverse covariance matrix using lasso penalty (L1 penalty) is used for partial correlation estimation and then, hub network graph is made by space (Sparse Partial Correlation Estimation) and available to be used for better decision making and developing a strategy in Marketing and Sales.