Integrating Crop Yield Prediction With Market Price Forecasting for Data-Driven Crop Selection in Kazakhstan

Abstract

Existing crop recommendation systems achieve high classification accuracy but rely solely on agronomic parameters, ignoring market dynamics that determine farmer profitability. This paper proposes an integrated framework combining machine learning-based yield prediction with time-series price forecasting for economically optimal crop recommendations. Using 14 years of data (2010-2023) across five major crops and 14 oblasts of Kazakhstan, coupled with ERA5 climate reanalysis and market price records, we compare multiple models for yield prediction (Linear Regression, Random Forest, XGBoost, LSTM) and price forecasting (ARIMA, Random Forest, LSTM, GRU). The best models-XGBoost for yield (RMSE 0.271 t/ha, R20.847) and GRU for price (MAPE 7.86%)-are integrated into a profitability scoring framework with Markowitz-inspired risk adjustment. Backtesting on 2021-2023 shows 23.7% higher simulated profit and 26.2% lower variance compared to yield-only recommendations, providing preliminary evidence that bridging agronomic suitability with market awareness can improve farmer decision support. The system is framed as a decision-support approximation; broader temporal validation is needed to confirm generalizability.

More by Nurzhamal Kashkimbayeva

• Evidence-Linked Measurement of Indirect Prompt Injection in Web Security Agents: Paired Trials and Application-Specific Evidence AuditsTechnologies · 2026
• Kazakhstan Agri Recommender — Data Collection PipelineZenodo · 2026
• Replication Data and Code for “Added Predictive Value and Temporal Transferability of MODIS and ERA5-Land Data for District-Level Wheat Yield Forecasting”Harvard Dataverse · 2026
• Multimodal Data Fusion for District-Scale Crop Yield Prediction in Northern KazakhstanIEEE International Conference on Smart Information Systems and Technologies · 2026