A Data‐Driven Framework for Crop Price Prediction Using ML, Statistical, and Hybrid Ensemble Models: Concepts, Challenges, and Applications

Manimegalai, R and Logendar, G and Srirengapriya, G (2026) A Data‐Driven Framework for Crop Price Prediction Using ML, Statistical, and Hybrid Ensemble Models: Concepts, Challenges, and Applications. In: Platform Engineering. Wiley, pp. 455-476. ISBN 9781394395910

Full text not available from this repository.

Abstract

Crop price prediction is one of the big areas of interest in agricultural economics, concerning farm planning and decision-making for the farmer, stakeholder, or policymaker. In this work, various machine learning models are developed to predict crop prices accurately using the historical dataset. The proposed methodology in this paper identifies the robust crop price prediction model. Models are built using random forest regression, long-short-term memory (LSTM), and statistical time series models such as autoregressive integrated moving average (ARIMA) and seasonal autoregressive integrated moving average (SARIMA). The models built using the above mentioned algorithms are compared on the basis of their accuracy and robustness. The proposed methodology involves gathering and pre-processing the dataset consisting of crop prices from historical data. Tasks such as feature engineering, model training, hyper-parameter tuning, ensembling, and evaluation are carried out in order to identify the robust model for crop prediction. Experimental results demonstrate the validity of the proposed methodology in terms of predicting crop prices accurately, with low error rate. The proposed approach contributes to the advancement in the methods of agricultural forecasting and improves insights into informed decisions in crop marketing and management. This will thereby offer several avenues for follow-up research, wherein, through considerations of weather condition, market dynamics, and crop-specific factors, it may be possible to make improved predictions. Random forest model achieved the highest accuracy, outperforming traditional statistical models. LSTM model effectively captured temporal dependencies, whereas XGBoost showed overfitting despite high training accuracy. The hybrid stacked ensemble model further improved prediction accuracy by combining multiple algorithms, enhancing crop price forecasting.

Item Type: Book Section
Subjects: Computer Science and Engineering > Machine Learning
Divisions: Computer Science and Engineering
Depositing User: Dr Krishnamurthy V
Date Deposited: 13 Aug 2026 09:40
Last Modified: 13 Aug 2026 09:40
URI: https://ir.psgitech.ac.in/id/eprint/1873

Actions (login required)

View Item
View Item