ESTIMASI BIAYA DESAIN UNTUK INDUSTRI MAKE-TO- ORDER DENGAN MENGGUNAKAN MACHINE LEARNING

• Main Author: Anenditya, Arsyeilla
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/80071
• Institution: Institut Teknologi Bandung

CV CSM is a manufacturing company with make-to-order production system. After receiving order, the company will first estimate the order cost before designing and processing the order. Due to high variety of products, designing has become a crucial step for the industry, including design costs. The aim of this study is to develop a cost estimation model using machine learning based on CAD and CAM data. Algorithms used in this study include, GBR, Random Forests, ANN, and SVR. The proposed model is then incorporated into a software prototype that can be operated by the problem owner for design cost estimation. Methodology used in this study is CRISP-DM. Steps include, business understanding, data understanding, data preparation, modelling, evaluation, and deployment. The proposed cost estimation model is a random forests model with an average R 2 score of 0,626 on testing data and 0,743 on training data. The application used to generate CAM data from CAD data was found to have weakness, such as incomplete extractions of machinable features in a component and some inaccurate strategy for machining processes. The proposed cost estimation model can be improved by adding more relevant data into training dataset and evaluating the data so that it’s still relevant to the company’s condition. Improving CAM data can be done by manually validating and modifying the CAM model generated by CAM application according to real case.