Application of the K-Nearest Neighbor Method (K-NN) for Wood Quality Selection as Raw Material for the Furniture Industry

Abstract

The furniture industry is a manufacturing sector that relies heavily on the quality of wood raw materials to produce products with strength, durability, aesthetic value, and high competitiveness. The process of determining wood quality is a crucial step because it affects the efficiency of raw material use, production costs, and the quality of the final product. However, the wood quality classification process in most industries is still carried out manually based on visual observation and expert experience. This approach has several weaknesses, including a high degree of subjectivity, inconsistency in the assessment process, and the relatively long time required when the number of pieces of wood to be inspected increases. Therefore, a classification method is needed that can provide objective, consistent, and accurate assessment results using a data-driven approach. This study aims to apply the K-Nearest Neighbor (K-NN) method to classify wood quality as a raw material for the furniture industry based on the wood's physical characteristics. The variables used in the study include wood diameter, moisture content, number of knots, presence of sapwood, tangential grain direction, and radial crack direction, all attributes that influence wood quality. Research data was obtained through direct observation and measurement of wood samples used as raw materials for production. Next, the data underwent a preprocessing stage, which included checking for completeness, data cleaning, categorical attribute transformation, and normalization using the Min-Max Normalization method to ensure all attributes were within the same value range and to prevent the dominance of certain attributes in the distance calculation process. The results showed that the K-Nearest Neighbor method was able to classify wood quality into high, medium, and low quality categories with a high level of accuracy and stable classification performance. Testing several k values ​​showed that selecting the optimal k value improved accuracy while reducing classification errors compared to k values ​​that were too small or too large. Furthermore, the K-NN-based classification process produced more objective decisions than conventional assessment methods because it was based on the similarity of historical data characteristics, rather than solely on the subjective judgment of observers. This research demonstrated that the application of the K-Nearest Neighbor method can be an effective alternative to support the wood quality selection process in the furniture industry. Implementation of this method is expected to increase the efficiency of the raw material inspection process, reduce the potential for classification errors, accelerate the decision-making process, and assist companies in selecting quality raw materials more consistently. Thus, the application of the K-NN method not only supports improving the quality of the furniture products produced, but also contributes to optimizing raw material management and increasing the competitiveness of the furniture industry as a whole.