[Paper Review] Predicting County Level Corn Yields Using Deep Long Short Term Memory Models

Zehui Jiang, Chao Liu+4 more|arXiv (Cornell University)|May 30, 2018

Energy Load and Power Forecasting12 references52 citations

TL;DR

The paper introduces a deep LSTM approach to predict county-level corn yields using cross-sectional time-series of yields and hourly weather data, showing promising predictive power in Iowa compared to survey-based methods.

ABSTRACT

Corn yield prediction is beneficial as it provides valuable information about production and prices prior the harvest. Publicly available high-quality corn yield prediction can help address emergent information asymmetry problems and in doing so improve price efficiency in futures markets. This paper is the first to employ Long Short-Term Memory (LSTM), a special form of Recurrent Neural Network (RNN) method to predict corn yields. A cross sectional time series of county-level corn yield and hourly weather data made the sample space large enough to use deep learning technics. LSTM is efficient in time series prediction with complex inner relations, which makes it suitable for this task. The empirical results from county level data in Iowa show promising predictive power relative to existing survey based methods.

Motivation & Objective

• Develop a deep learning approach (LSTM) to predict county-level corn yields.
• Leverage cross-sectional time-series data combining county yields with hourly weather information.
• Evaluate predictive power relative to traditional survey-based yield prediction methods.
• Demonstrate feasibility and potential benefits for price discovery and market efficiency in futures markets.
• Focus on Iowa county data to illustrate the method’s applicability.

Proposed method

• Apply Long Short-Term Memory (LSTM) networks to time-series yield prediction.
• Construct a cross-sectional time-series dataset linking county yields with hourly weather data.
• Compare LSTM predictions against existing survey-based prediction methods.
• Explain LSTM’s capability to model complex temporal relationships in agricultural data.
• Use county-level data from Iowa to assess predictive performance.

Experimental results

Research questions

• RQ1 Can LSTM models accurately predict county-level corn yields from county yields and hourly weather data?
• RQ2 How does LSTM predictive performance compare to traditional survey-based yield prediction methods?
• RQ3 Does the LSTM approach provide effective yield forecasts for Iowa counties?

Key findings

• LSTM demonstrates promising predictive power relative to survey-based methods.
• Empirical results based on Iowa county data support the effectiveness of the deep learning approach.
• The use of hourly weather data enables handling a large sample space suitable for deep learning.
• The study is the first to apply LSTM to county-level corn yield prediction.