This paper focuses on the issue of remote construction sites jobs detection and their progress assessing. Overseeing remote system for the tasks progress control is proposed. The proximity sensors were mounted in the uncompleted building and workers’ pattern of movement was tracked. Compering data from the sensors and the progress of works allowed to conduct the affinity analysis that provide the association rules for finding the correct locations of the tasks performed each day. This study shows the process of finding proper values for rules selection that ensure the correct indication of place where tasks were accomplished on given day and propose such values for typical block of flats. Two approaches to association analysis are proposed with the intention of highlighting differences between rules found with both high and low levels of support within dataset. During data analysis numerous association rules are found so strict requirements considering supports, confidence and correlation are introduced. Proposed rules are assessed and the found antecedents in the rules of the high confidence are believed to be critical bottlenecks for production process. For future endeavors, the placement of sensors is proposed based on the positive predecessors identified in the discovered rules. These rules indicate that fewer detection points are actually required to provide accurate data. Finally in paper future scope of needed tests and experiments is mentioned in order to develop the ability to utilize data mined from construction sites.