Decision Support for SWEA Inspections

Abstract: Nurturing a good working environment is beneficial not only in the short term for keeping employees happy, but especially important for work-life longevity. Poor working conditions can severely lower the retirement age depending on the line of business. The Swedish Working Environment Authority (SWEA) is a governmental institution that regulates and enforces work environment laws. Currently, SWEA has issues in justifying which workplaces to choose for inspection. Therefore, they are interested in implementing statistical methods to help them make decisions about which workplaces to inspect for work environment issues. This thesis implements a prototype logistic regression model to classify workplaces as suitable to visit or not. There are two definitions for which workplaces that are visit suitable, either due to the number of deficiencies a workplace had (median group) or due to the severity of an incident at the workplace (severe group). Results are presented for both. Considering a subset of the workplaces that have been given the highest probability of being visit suitable, the prototype performed slightly better than random in the median group and significantly better than random in the severe group.