Festivaalikävijöiden ryhmittely k-means- algoritmin avulla

Festivals and events call for substantial investments. To meet business goals, event managers aim to understand their customers’ preferences to enable the design and production of services to meet needs and expectations, which in turn helps generate revenue. Customer segmentation is an established approach to this problem. Previous research leverages multivariate survey data and cluster analysis to provide an understanding of customer preferences. The primary objective of this thesis is to apply this methodology to transaction data from a festival. In this thesis, transaction data is processed to represent customer preferences based on their purchasing behaviour. The dataset used is the full transaction data of a large outdoor festival over multiple days. Clustering with the k-means algorithm is applied to assign festivalgoers into customer segments based on their preferences. Clustering performance is evaluated with established internal metrics that are defined by the properties of the resulting clusters. Two different hierarchical levels are formed from the data, and clustering parameters are adjusted based on their performance. The clustering results and parameters are defined by the best performance achieved with the chosen k-means methodology. The resulting customer segments are evaluated in regards to their usefulness in event management. Established properties for efficient customer segmentation are leveraged for this analysis. The methodology helps define meaningful customer segments. The results are used to gain useful insight into customer behaviour to design service production appropriately. To increase usefulness, future research should explore temporal and monetary aspects of the data that are not utilized in the applied methodology. The generalizability of the resulting segments has not yet been validated, as the results of this thesis are limited in scope by the dataset from a single festival.