This master thesis aims to demonstrate, how the dimensioning of a warehouse may be supported by doing an automated analysis of inbound and outbound data. The warehouse planning process is mostly based on key figures that each considers the data of the entire product range. In the future, this automated analysis should be done by calculating key figures for product groups, which are derived from the product range automatically as special handling operations are needed. An algorithm is used to create various clusters which contain similar products. Furthermore, the topic of data science is discussed as this field deals with the theoretical foundations of data mining. A cluster analysis is realized to divide a real dataset into different clusters, each with similar objects. The implementation of a cluster analysis is based on developing a clustering algorithm, which consists of three core elements: specific key performance indicators of the data, an adequate metric and a suitable clustering method. An appropriate combination of these three influencing factors may yield into a valuable outcome performed by the clustering algorithm. The presented cluster analysis in this master thesis has been executed for products which appear temporally concentrated and have also a high sales volume. The evaluation of the clustering algorithm was elaborated in order to verify, how well products with a temporary high volume have been combined into one cluster. In consequence of selecting the required cluster, a classification problem appears. The numbers of true positives, true negatives, in addition to false positives and false negatives may be evaluated through a confusion matrix. Common ratios like the sensitivity and the specificity were derived in order to compare different clustering results. Additionally, a statistical test has given insight, as to which likelihood the algorithm’s ability to recognize clusters applies to the entire dataset. The results show that products with the characteristic of a temporarily concentrated and high sales volume are found in one cluster. The precise suitability of the developed clustering algorithm for the given data set has been proven. As this represents a valuable gain in information for warehouse dimensioning, the clustering algorithm may be used in order to cluster inbound and outbound data of warehouses.