Long term and short term forecasting of horticultural produce based on the LSTM network model.

Forecasting the price of agricultural produce helps grower decide planting, harvesting, and trading time. Price forecasting of crops has garnered many researchers' attention, hence plenty of forecasting models already exist in literature. Price forecasting of horticultural products is still a lesser explored area for researchers. Pricing strategy of vegetables do not follow the same strategies as that of crops. Due to seasonality and short lifetime, maintenance of vegetables differ from other agricultural products. As horticultural products are generated in a very short time after plantation, close forecasting of price may provide farmers more profit by choosing vegetables for plantation and guiding the appropriate harvesting time. In this paper, we have analyzed the performance of some machine learning and statistical models in this front. The error metrics like Root Mean Square Error(RMSE), Mean Absolute Error(MAE), and Mean Absolute Percentage Error (MAPE) have been used to study the performance of the models. We have proposed a Long Short-Term Memory(LSTM) based model for long-term price forecasting of vegetables like cabbage, Cauliflower, and Brinjal for some Indian markets. Friedman test and Wilcoxon signed rank test are used to analyze the employed models' similarity and dissimilarity. The experiment results indicate that the proposed model outshines other models. A short-term price forecasting model has also been experimented with in this paper. [ABSTRACT FROM AUTHOR]