Smart Crop Yield Forecasting Through Hybrid Attention-LSTM Deep Learning

Agriculture is an important profession that is dependent on the weather and the amount of precipitation that occurs across the world. In order to predict crop yields in advance, the purpose of this article is to make use of data on climate, soil, and temperature. A classification-based technique for agricultural output prediction is proposed in this paper. The methodology makes use of Long Short-Term Memory (LSTM) in conjunction with an Attention Mechanism method. It is the department of Economics and Statistics within the Karnataka government that is responsible for collecting the manual data. The Department of Economics and Statistics provided the information that was utilized in this method. The three crops that were employed were jowar, rice, and ragi. A method known as linear interpolation is used in order to complete the dataset by filling in the values that are absent or null. Both the Correlation-based Feature Selection Algorithm (CBFA) and the Variance Inflation Factor Algorithm (VIF) may benefit from the feature selection technique since it assists them in selecting and removing groupings of characteristics that are connected to one another. Accuracy, R2, Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) are the metrics that we use in order to evaluate the functional effectiveness of the model. The LSTM model that was presented produces outcomes with evaluation metrics such as accuracy, R2, MAE, MSE, and RMSE values that are about 99.10%, 0.44, 0.132, and 0.233, respectively according to the suggested model.