Indonesian Journal of Electrical Engineering and Informatics (IJEEI) Vol. 10, No. 4, December 2022, pp. 825~836 ISSN: 2089-3272, DOI: 10.52549/ijeei.v10i4.3946  825 Journal homepage: http://section.iaesonline.com/index.php/IJEEI/index Forecasting and Clustering of Cassava Price by Machine Learning (A study of Cassava prices in Thailand) Sayan Tepdang 1 , Ratthakorn Ponprasert 2 1,2 Faculty of Business Administration and Information Technology, Rajamangala University of Technology Tawan-Ok, Chakrabongse Bhuvanarth Campus, Bangkok, Thailand Article Info ABSTRACT Article history: Received Jun 23, 2022 Revised Nov 15, 2022 Accepted Nov 28, 2022 Forecasting and clustering the price of cassava is essential for agriculture, but the difficult part of forecasting is price fluctuation, in which the fact of prices is going up and down and be changed monthly. The paper proposes to forecast the price of Cassava by machine learning. The process had been calculated by the price of Cassava from January 2005 to February 2022, which has been collected for 17 years by the Office of Agricultural Economics, Ministry of Agriculture, and Cooperatives. The research on forecasting found that the method of Support Vector Machine including using add-on feature with Garlic Price and Potato Price showing the Root Mean Squared Error (RMSE) with the lowest point as of 0.10. If comparing to the Conventional method with the equal database. The result shows that the proposed method demonstrates the value of the Mean Absolute Percentage Error (MAPE) as 3.35%, it displays more effectively as 0.61%. For the final process of clustering the price by analyzing with K-mean, the result came up with a peak pricing period in December of 14.08%. Subsequently, the agricultures would apply the research result to implement their planting plan for profit-making. Keywords: Machine Learning Data Mining Forecasting of Cassava Price Data Analysis Knowledge Discovery in Data Copyright © 2022 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Sayan Tepdang Faculty of Business Administration and Information Technology, Rajamangala University of Technology Tawan-Ok, Chakrabongse Bhuvanarth Campus, Bangkok, Thailand Email: sayan_te@rmutto.ac. 1. INTRODUCTION Cassava is plant-based, which is a high-ranking in the top five subordinating to Corn, Wheat, Glutinous Rice, and Potato. It also has a high quantity of carbohydrate as well as provide various benefits. For instance, the food - as the consumable product for human and husbandry, the fuel – as the main ingredient for the Ethanol production process, and the manufacturing – as the component of the Alcohol and Apparel production process, etc. In 2019, the global production of Cassava came up to 309 million tons. Africa is the largest source of global products leading to Asia, America, and Oceania continent. For the country, Nigeria is the top producer leading to Congo, Thailand, Ghana, Brazil, and Indonesia. According to Thailand by 2020, the country has a cultivated land of approximately 3.5 million acres located in Nakhonratchasima, Kampangpetch, Chaiyaphum, and Kanchanaburi, respectively[1]. The cost of planting cassava in Thailand is approximately 5,974.78 THB for 0.39 acre. The average cost from 2011 to 2021 is 3.398 tons for 0.39 acre [2]. According to the methodology of the research, there are 2 approaches; 1) the Mathematical and Statistical approach and 2) the Machine Learning approach. Firstly, the Autoregressive Integrated Moving Average (ARIMA) method in terms of the Mathematic and Statistic, which is a showcase of predication, is implied in the related research papers to forecast the price of agricultural products. For instance, aromatic coconut [3], field crops [4], litopenaeus vannamei [5], and mango export volumes [6]. In addition, the method also implied forecasting for the Cassava product, for example: fresh cassava root buying prices and cassava chip selling prices [7], export volumes of cassava [8], and cassava price [9] ,[10]. As a result, the method is a fast approach for processing and forecasting, meanwhile, both variable and suitable methods remain an