Exploring the representativeness of the M5 competition data

The main objective of the M5 competition, which focused on forecasting hierarchical unit sales Walmart, was to evaluate accuracy and uncertainty methods in field identify best practices highlight their practical implications. However, can findings competition be generalized exploited by retail firms better support decisions operation? This depends extent data is sufficiently similar retailers operating different regions selling product types considering marketing strategies. To answer this question, we analyze characteristics time series compare them with those two grocery retailers, namely Corporación Favorita a major Greek supermarket chain, using feature spaces. Our results suggest only minor discrepancies between examined sets, supporting representativeness data.