Jadwal Sholat

Memuat jadwal sholat…

Ilmu Komputer & AI editorial

Open AccessOA2023

Predicting Agriculture Yields Based on Machine Learning Using Regression and Deep Learning

This study evaluates decision tree, random forest, XGBoost, CNN, and LSTM for crop yield prediction. Random forest achieved 98.96% accuracy, while CNN showed the lowest loss (0.00060), demonstrating the potential of machine learning to support agricultural decision-making.
Priyank Sharma; Pankaj Dadheech; Nagender Aneja; Sandhya Aneja· IEEE Access· 2023· DOI 10.1109/ACCESS.2023.3321861

The core problem

Agriculture is a cornerstone of India's economy, supporting a large population and contributing significantly to livelihoods. However, food security is threatened by rapid population growth, which increases demand for food while arable land remains limited. Farmers must therefore enhance productivity on existing land. Crop yield prediction offers a data-driven solution to assist farmers in making informed decisions about crop selection and management throughout the growing season. This paper aims to predict crop yield using variables such as rainfall, crop type, meteorological conditions, area, production, and yield. By leveraging machine learning (ML) and deep learning (DL) techniques, the study seeks to reduce agricultural output losses and support sustainable agriculture. The primary objective is to compare the performance of various ML and DL algorithms—decision tree, random forest, XGBoost, convolutional neural network (CNN), and long short-term memory (LSTM) network—for crop yield prediction, evaluating them on metrics including accuracy, root mean square error (RMSE), mean square error (MSE), mean absolute error (MAE), standard deviation, and loss.

Innovation

The study utilizes a dataset comprising rainfall, crop, meteorological conditions, area, production, and yield variables. The goal is to predict crop yield based on these features. Five algorithms are employed: three regression-based ML methods (decision tree, random forest, and XGBoost) and two DL methods (CNN and LSTM). The models are trained and tested on the dataset, and their performance is evaluated using the following metrics:

- **Accuracy**: For regression, accuracy is interpreted as the percentage of predictions within a certain tolerance or as \(R^2\) score; the paper reports a maximum accuracy of 98.96% for random forest.
- **Mean Absolute Error (MAE)**: \(\text{MAE} = \frac{1}{n} \sum_{i=1}^{n} |y_i - \hat{y}_i|\)
- **Root Mean Square Error (RMSE)**: \(\text{RMSE} = \sqrt{\frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2}\)
- **Mean Square Error (MSE)**: \(\text{MSE} = \frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2\)
- **Standard Deviation**: \(\sigma = \sqrt{\frac{1}{n} \sum_{i=1}^{n} (y_i - \mu)^2}\)
- **Loss**: For DL models, training loss is monitored; CNN achieved a minimum loss of 0.00060.

The data is split into training and testing sets, and models are tuned for optimal performance. The comparison aims to identify the most effective approach for yield prediction.

Introduction
Agriculture is a cornerstone of India's economy, supporting a large population and contributing significantly to livelihoods. However, food security is threatened by rapid population growth, which increases demand for food while arable land remains limited. Farmers must therefore enhance productivity on existing land. Crop yield prediction offers a data-driven solution to assist farmers in making informed decisions about crop selection and management throughout the growing season. This paper aims to predict crop yield using variables such as rainfall, crop type, meteorological conditions, area, production, and yield. By leveraging machine learning (ML) and deep learning (DL) techniques, the study seeks to reduce agricultural output losses and support sustainable agriculture. The primary objective is to compare the performance of various ML and DL algorithms—decision tree, random forest, XGBoost, convolutional neural network (CNN), and long short-term memory (LSTM) network—for crop yield prediction, evaluating them on metrics including accuracy, root mean square error (RMSE), mean square error (MSE), mean absolute error (MAE), standard deviation, and loss.

Why it matters

The results demonstrate that both ensemble regression (random forest) and deep learning (CNN) are promising for crop yield prediction. Random forest's high accuracy and low error metrics make it a robust choice for regression tasks, as it effectively handles non-linear relationships and interactions among features. CNN's low loss indicates its capacity to learn hierarchical representations from data, which is beneficial when dealing with complex patterns such as those in meteorological and rainfall data. However, the study does not provide a direct comparison of CNN's accuracy or error metrics beyond loss, limiting a full assessment. The authors note that other ML and DL methods fell short compared to random forest and CNN. The findings are analyzed using RMSE to understand how the model's errors compare to other methods. Overall, the study contributes a comparative evaluation of ML and DL techniques for agricultural yield prediction, highlighting the potential of these models to support farmers in decision-making. Future work could explore hybrid models, incorporate more diverse data sources, and validate on larger datasets to enhance generalizability.

Who should read this

Pembaca Research Digest dan praktisi rumpun ini.

Opening member content…