This study focuses on delivery performance in postal services by quantitatively exploring data of operational measurements. The goal is to lay out the most important factors that affect delivery performance. The data is provided by Postal Service reporting team, Posti Group Oyj.

First, the thesis presents the theoretical background of machine learning usage and algorithms that were used to get information about factors that affect the most on delivery performance rate. The data of this study was analysed using two regression models (Generalized Linear Regression and Random Forest Regression models) in order to have models with good interpretation possibilities. The models were evaluated by RMSE and R2 error metrics. Next, the thesis describes the factors that have been observed to have affect the work performance in the past literature: training, information and communication technologies, infrastructure level, automation level, supply chain complexity, health, and work-related factors (age, type of contract).

The identifies key factors as: sickness rate, difference of planned and actual volume, difference of planned and actual number of employees, indoor efficiency, percentage of overwork hours, route master application rate. Both regression model types applied in this study produced near to similar results. However, the models had low percentage of described variance and relatively high RMSE that indicated the need of further exploring other factors that might have an impact on the delivery performance rate.