This paper discusses a case study in which publicly available data of a rail freight
transportation firm has been gathered, cleansed, and analyzed in order to: (i) describe the
data using statistical indicators and graphs; (ii) identify patterns regarding several Key
Performance Indicators; (iii) obtain forecasts on the future evolution of these indicators; and
(iv) use the identified patterns and the generated forecasts to propose customized insurance
products that reflect the current and future freight transportation activity. The paper
illustrates the different methodological steps required during the extraction and cleansing of
the data --which required the development of Python scripts--, the use of time series analysis
for obtaining reliable forecasts, and the use of machine learning models for designing
customized insurance coverage from the identified patterns and predicted values.
