Analysis of the Environmental Impact of Food Production in Indonesia using Machine Learning Models Based on FAO Data

Aug 6, 2026

Upload by Violet

Food production in Indonesia underpins national food security while driving substantial environmental pressures through greenhouse gas emissions, land use change, and water consumption. This study builds a machine learning framework using Food and Agriculture Organization (FAO) data to examine links between major food commodities and ecological footprints. FAO-derived datasets are cleaned and transformed, then clustered to group commodities by impact and modeled with XGBoost to predict emissions, achieving an RMSE of 3.07 MtCO₂e. Results identify rice, palm oil, and root crops as high-impact contributors, with rice delivering the largest emissions from 1960–2013. The work supports data-driven sustainability strategies for Indonesia’s food sector.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Introduction

Literature Review

Analysis of the Environmental Impact of Food Production in Indonesia using Machine Learning Models Based on FAO Data Oktaviana Maria Noni Imil1, I Kadek Agus Wahyu Raharja2*, Gde Wikan Pradnya Dana3 1,2,3Computer Engineering Program, Faculty of Engineering and Planning, Warmadewa University ABSTRACT: Food production in Indonesia is a critical pillar of national food security but has significant environmental impacts, including greenhouse gas emissions, land use, and water consumption. This study analyzes the environmental impact of major food production in Indonesia using a Machine Learning (ML) approach based on data from the Food and Agriculture Organization (FAO) . Data from FAO.csv, Food_Production.csv, and total_population_reform.csv were processed to identify relationships between food commodities and their ecological footprints. ML techniques, such as clustering and the XGBoost model, were employed to group commodities based on environmental impact and predict total emissions with an RMSE of 3.07 MtCO₂e. Results indicate that commodities like rice, palm oil, and root crops have significant environmental impacts, with rice contributing the highest emissions at 374.684 MtCO₂e (1960–2013). This study provides data-driven strategies to support the sustainability of Indonesia’s food sector, aligned with green technology principles, through visualizations of supply and emissions for the top 10 commodities. KEYWORDS: Food production, environmental impact, Machine Learning, sustainability, FAO data I. INTRODUCTION Indonesia, as an agrarian nation with a population exceeding 270 million, faces significant challenges in ensuring food security while preserving environmental sustainability. Food production, such as rice, palm oil, and root crops, dominates the national agricultural sector but also contributes significantly to climate change through greenhouse gas (GHG) emissions, land degradation, and intensive water and land use [1] [2] [3] . Agriculture accounts for over 14% of national GHG emissions, with the food sector being a major contributor through activities like land preparation, fertilizer use, and transportation of harvested products [4] . Globally, food production contributes approximately 26% to total GHG emissions, with a significant proportion originating from developing countries like Indonesia [2] . Rice, a staple commodity in Indonesia, has a high carbon footprint due to wetland cultivation methods that produce methane, a greenhouse gas 25 times more potent than carbon dioxide [5] . Additionally, the expansion of palm oil plantations has led to widespread deforestation, increasing carbon emissions and reducing biodiversity [6] . These challenges are exacerbated by rising food demand due to population growth, projected to reach 300 million by 2030 [7] . Previous studies have highlighted the relationship between food production and environmental impacts, but most focus on global or regional scales, often overlooking Indonesia’s specific context [2] [3] . Research in Indonesia is frequently limited to qualitative analyses or incomplete datasets, hindering the development of data-driven mitigation strategies [8] . With technological advancements, Machine Learning (ML) approaches offer solutions for analyzing complex datasets like those provided by FAO, enabling accurate predictions and identification of environmental impact patterns [9] . The use ofFAO data in this study provides an advantage due to its extensive coverage, including historical food production, emissions, and resource use data from 1961 to 2013 [1] . This data enables rich longitudinal analysis to understand long-term trends and relationships between variables. However, challenges such as inconsistent formatsand missing values require careful preprocessing, addressed in this study through data cleaning and transformation techniques [10] . Machine Learning approaches, particularly the XGBoost model and clustering, were chosen f

FAQs

What environmental factors does the study analyze in Indonesia’s food production?

It focuses on greenhouse gas emissions, land use impacts, and water consumption associated with major food commodities.

Which machine learning methods are used to model and group commodities?

The study uses clustering to group commodities by environmental impact characteristics and XGBoost to predict total emissions.

What is the model’s prediction performance and key findings?

The XGBoost-based prediction achieves an RMSE of 3.07 MtCO₂e. Rice, palm oil, and root crops show significant impacts, with rice contributing the highest emissions between 1960 and 2013.

Aug 6, 2026

Upload by Violet