COMPREHENSIVE ANALYSIS AND MACHINE LEARNING FRAMEWORKS FOR AGRICULTURAL PRICE STABILITY IN INDIA

Abstract

Agriculture significantly contributes to the Indian economy as it provides millions of farmers with job security and helps ensure food security. However, agricultural commodity prices in India can be very unpredictable and unstable due to various factors, including crop seasonality, weather patterns, transportation issues, government policies, and the fluctuation of demand. Therefore, this makes it challenging for farmers, traders, and policymakers to make educated decisions about planting crops, how to store them, and when to sell them. The project will create a predictive analytics framework to forecast the prices of agricultural commodities based on machine learning methods and time series analysis. Data from multiple states and mandis (markets) across India for various crops (such as wheat, rice, onions, potatoes and tomatoes) were collected and compiled into a dataset so that they could all be assessed and analysed together. This study has used an extensive data preprocessing pipeline to enhance the quality of the data by cleaning it, standardising it, correcting errors and inconsistencies, detecting outliers using Interquartile Ranges (IQR) and creating additional features. To capture how prices behaved over time, several lagged (i.e., historical) and rolling statistical features were created, including lagged prices, rolling averages and rolling volatility measures. Once the final dataset had been created, several different predicting models (e.g., ARIMA, linear regression, random forest, gradient boosting and XGBoost) were developed to forecast future agricultural commodity prices. The predicting models were evaluated using mean absolute error (MAE), R2 and Pearson correlation coefficients (R) to determine how accurately they were able to predict and how reliable they were. Overall, the experimental data indicated that machine learning techniques generally outperformed traditional statistical methods when applied to the data set chosen for this project. As such, of all methods tested, the Linear Regression technique produced the most accurate estimate of the future price of Wheat (Deshi) with the lowest average error as well as the greatest correlation between predicted and actual prices. Conversely, ARIMA was not able to effectively model the non-linear fluctuations in market price. In addition to the forecasting objectives of this project, prescriptive analytics were applied using the developed forecasting techniques through the use of market-wise price comparisons and sell-or-hold recommendations to assist agricultural producers/market participants in making more informed and practical decisions regarding the timing of their sales. Lastly, clustering analysis will allow for the Ⅳ (All Multidisciplinary Journal,2024)(All Multidisciplinary Journal, n.d.)identification of "like" markets characterized by similar pricing patterns and volatility profiles. This research illustrates how utilizing a data-driven forecasting system can lead to better decision making in agriculture by providing farmers, traders, and policy makers with valuable information about the future of agricultural markets in India. Intelligent agricultural advisory systems developed on this framework will enable better price stability, lower levels of uncertainty, and a more efficient functioning of agricultural markets within India. Agmarket.net Under Market Research and Information Network (MRIN) sub scheme of Integrated Scheme for Agricultural Marketing (ISAM) an ICT based Agricultural Marketing Information Network (AGMARKNET) was launched in March 2000 to link important agricultural produce markets spread all over the country and the State Agriculture Marketing Boards and Directorates. More than 4000 APMCs are on boarded on the AGMARKNET portal and more than 2700 APMCs are regularly reporting data on the portal. More than 450 commodities and 2000 varieties are covered under the scheme. The Directorate of Marketing and Inspection, Ministry of Agriculture & Farmers Welfare, Govt of India is implementing the scheme in association with the State Agricultural Marketing Boards/Directorates and APMCs. Objectives of the Scheme To establish a nation-wide information network for speedy collection and dissemination of market information and data for its efficient and timely utilization. To facilitate collection and dissemination of information related to better price realization and market access by the farmers. This would cover: ● Market related information such as market fee, market charges, costs, method of sale, payment, weighment, handling, market functionaries, development programmes, market laws, dispute settlement mechanism, composition of market committees, income and expenditure, etc. ● Price-related information such as minimum, maximum and modal prices of varieties and qualities transacted, total arrivals and dispatches with destination, marketing costs and margins, etc. ● Infrastructure related information comprising facilities and services available to the farmers with regard to storage and warehousing, cold storage, direct markets, grading, re-handling and Ⅴ repacking etc. ● Market requirement related information covering accepted standards and grades, labeling, sanitary and phyto-sanitary requirements, pledge finance, marketing credit and new opportunities available in respect of better marketing. To improve efficiency in agricultural marketing through regular training and education for traders/major specific farmers in their local language. To provide assistance for marketing research to generate market information for its dissemination to farmers and other market functionaries at grass root level to create an ambience of good marketing practices in the country. For information relating to the Schemes in respect of agricultural marketing implemented by the Govt. (Central and States) and their financial assistance. Once the farm produce is standardized and labeled, backed by quality certification, it can be directly offered for sale in spot exchange in national and international markets.((Directorate of Marketing and Inspection, 2026)