Vliv komplementů na dynamickou cenotvorbu

This thesis addresses the topic of dynamic pricing, or more specifically finding optimal prices of products on e-shop utilizing demand curves approximated by machine learning models. The thesis examines the effect of complementary products that customers often buy together and aims to extend these models with new predictors relating to these products and try to reduce the prediction error this way thus improving the optimized price output. The thesis contains an R implementation of market basket analysis to identify these association rules in the transaction history as well as instructions on how to use it. Afterwards it describes a high-level architecture of the current solution and specifies how the new predictors are added into it. Finally, it provides a comparison between the original and extended models followed by a discussion of the results that supports the original hypothesis that dynamic pricing models can be made more accurate by adding the effect of complements.