Inductive learning methodology using OSHA construction accident and citation data.

摘要

This research idea started with the simple concept that there are causal factors, if present together, leading to a high probability of a construction accident. To explore and define those factors we solicited accident information through several sources. All of these sources did not provide us with adequate encoded data that we could work with. However, detailed accident investigation reports and their associated citations had been published by OSHA over its website and were available. These reports have provided us with a comprehensive set of data that we can organize and manipulate. Out of these data mine, relevant factors were extracted and processed to obtain the information that contributed to this research.; A simple relational database that can organize, control, process, and more importantly, update data, was created using the Microsoft Office Excel program. A systematic process was used in this research to eliminate the superfluous data and to identify mutually exclusive independent variables to arrive at an overall the data set. The refinement process allowed us to focus on critical and effective variables, and to build and explore the relationship among the variables.; Data was grouped and sorted out. Values of each variable were identified and their frequency occurrence was counted. Simple analytical tools were utilized to achieve inductive learning to explore the relationships among accident and citation variables levels, either separate or combined. These analytical tools used were: (1) Frequency analysis (bar chart, pie chart) and probability and frequency-based techniques; (2) Simple counting and utilization of mode-based relationships; and (3) Probabilistic relationships (probability of joint occurrences) to explore relationships between variables.; Certain relationships emerged out of this research that will enhance the safety practices and awareness in the construction industry, pinpointing those causal factors in order to eliminate them or at least to pay more attention to their presence.; In conclusion this research introduced a new methodology to study and understand construction accidents. In an industry that is characterized by a high accident rate and poor safety practices, this study will pave the road for further studies with more accurate results that will enhance safety records of the construction industry.