Identifying the way in which the enterprises of related businesses locates in proximity occur - the coagglomeration of industries - is a key solution to understand the geographic concentration mechanisms of industries. Many research in geography and spatial economics have been addressing this issue to explain and theorize such mechanism. This paper proposes a new statistical approach that allows extracting coagglomeration patterns of industries from available national datasets. The approach takes two steps; the first step finds spatial clusters of each industry using the false discovery rate-controlling statistical test, and the second step searches for colocation relationships among industries through the frequent pattern mining of detected cluster location and the Monte Carlo simulation. This approach identifies coagglomeration patterns of industries. The proposed new method is applied to the 500-meter grid data of the 2012 Economic Census for Business Frame of Japan, to check its applicability and validity.