Clustering Nigerian male traditional fashion designs using weighted similarity and tabu search for feature selection, and linear minimum cost flow clustering Department:Industrial and Production Engineering By: El_Zarh007 Project ID: 7919 Rating: (5.0) votes: 2 Price:₦5000 Get the Complete MaterialAbstractFashion is an important part of our daily lives, and the fashion industry has seen rapid growth in recent years, especially in this age of Artificial Intelligence. The downside is that the Nigerian traditional fashion industry is still trailing in adapting to these changes as traditional fashion designs cannot be simply described by standard semantic attributes, unlike foreign clothing styles. The aim of this study is to develop a machine-learning model that can automatically group different male traditional clothing designs based on their semantic features to establish attribute standards. This project exploited the power of the metaheuristic search technique, weighted similarity, and Tabu Search (WS-TSA) for selecting the five most relevant features from the original sample set of attributes and applied Dijkstra’s Shortest Path algorithm, which is a variant of the Linear Minimum Cost Flow Network algorithm (LMCFN), to carry out the actual clustering of the dataset. A total of 100 data samples were collected at random from various e-commerce and social platforms. Features obtained were weighted using Statistical or Shannon entropy and reduced using the Weighted similarity and Tabu search methods before clustering with the LMCFN model. Finally, the points assigned to clusters generated were validated using the silhouette index to determine how well the data points fit into each cluster. The result of the tested data generated the top 5 attributes with their corresponding weights as (Number of Pieces, Sleeve length, Design Type, Cloth Length, Design Type for Outer Garment), and (0.57, 0.77, 0.44, 1.00, 0.00), respectively. The clustering model generated four distinct clusters with counts (29, 12, 28, 30). Breaking it down, the clustering model had about 72% expected performance using the silhouette score. ...Preview Download Preview +Other Industrial and Production Engineering project topics and materials you might be interested in»A fuzzy-analytic hierarchy process for measuring flood resilience of a commercial property»Clustering Nigerian male traditional fashion designs using weighted similarity and tabu search for feature selection, and linear minimum cost flow clustering