Determining consumer demand patterns for production planning using a data mining approach

The bread industry faces significant risks of losses in case of excess inventory. The initial stage in the K-Means Clustering Algorithm involves forming two clusters: C1 for slow-moving product data and C2 for fast-moving. Clustering products using the K-Means Algorithm resulted in Group 1 as slow-moving products with 44 types of items and Group 2 as fast-moving products with 15 types of items. It can be concluded that the bakery is experiencing losses due to an excess of overstocked products. After categorizing data into slow-moving and fast-moving groups, the subsequent phase involves employing the FP (Frequent Pattern)-Growth association rule algorithm to recognize consumer purchasing patterns. This algorithm aims to uncover relationships between items in a dataset and assess the probability of a person purchasing bread concurrently. By establishing a minimum support of 3% and a minimum confidence level of 30%, a total of 13 rules were generated, meeting the criteria for strong association rules. With this data, the store owner can specifically enhance inventory planning for fast-moving products by analyzing demand data and market trends. For slow-moving products, the store owner can adjust item placement or create product bundling with best seller items.