ABSTRACT

Spices are conventional aromatic vegetables mainly utilized for flavouring of food. Among these, chilli (Capsicum annuum), is one of most important spice used around the world. The cultivation and trade of spices, particularly chilli, play a significant role in global culinary practices, with India is major hub in this domain. Renowned as the "Spice Bowl of the World," India's abundant production, consumption, and exportation of spices underscore its pivotal position in the industry. However, the volatility inherent in horticultural markets, exacerbated by natural calamities, necessitates robust forecasting mechanisms to empower farmers to make informed decisions. Recognizing this need, a study was conducted to predict the price of dry chilli in the Bydagi market of Haveri district, Karnataka. Leveraging secondary data sourced from Agmarknet spanning from 2000 to 2022, supervised machine learning techniques were employed, specifically employing Python within a Jupyter notebook, with Artificial Neural Network (ANN), Recurrent Neural Network (RNN), Long Short Term Memory Neural Network (LSTM), Random Forest (RF), and Decision Tree (DT), models scrutinized. The findings underscored the efficacy of the LSTM exhibit superior than ANN, and RNN and RF exhibiting superior performance compared to the DT. The testing R2 values for deep learning models are ANN (0.58), RNN (0.82), LSTM (0.93). Similarly, for Machine learning models RF (0.91), and DT (0.85) and other metrices are also used for comparison of models. This research culminates in a forecast model poised to offer tangible benefits to dry chilli farmers, furnishing them with invaluable insights to navigate the dynamic Agricultural landscape. By leveraging advanced analytical techniques, stakeholders can mitigate risks, optimize resource allocation, and bolster resilience in the face of market fluctuations, thereby fostering sustainability and prosperity within the spice industry.