As is well known, resident engineers on site might plan a heuristically deterministic construction schedule by PERT/CPM program. Since it is boring and time-consuming to gather, edit and input field data into the PERT/CPM program, however, there are very few opportunities to revise the construction schedule on daily works. This paper proposed an easy-to-use and -learning method to plan and revise construction schedule and to do ensemble forecasting construction schedule. First, this paper outlines this study and analysis methods to calculate daily real operational hour and rate based on 3-axis acceleration excited by machine operation. The 3-axis acceleration response values could be collected by sensor built in on-board smartphone for construction machine. In addition, the paper explains methods applied to the machine hour data and rate, to determine shape parameters of Beta distribution, to execute bagging based on samples from the Beta distribution, and to conduct ensemble forecasting of construction schedule. Finally, this paper discusses lesson learned from the empirical study and discussion of predictability of the ensemble forecasting of construction schedule.