Coffee Price Prediction: An Application of CNN-BLSTM Neural Networks

Abstract

Coffee is one of the world's most popular beverages, and its production and demand have been steadily increasing in recent years. In 2020/21, worldwide coffee output hit 174.5 million bags, according to the International Coffee Organization. coffee year, which is a 1.9% increase from the previous year. The demand for coffee is driven by various factors, including changing consumer preferences, economic conditions, and demographic trends. In particular, the growing popularity of specialty coffee and the increasing consumption of coffee in emerging economies have contributed to the growth in demand. However, the coffee market has also faced challenges such as climate change, which can affect coffee production by altering the growing conditions, and the COVID-19 pandemic, which has disrupted supply chains and caused fluctuations in prices.In terms of regional When it comes to coffee output, Brazil leads the globe, followed by Vietnam, Colombia, and Indonesia. These countries collectively account for more than 60% of global coffee production. The United States, Germany, and Japan are the largest importers of coffee.Overall, coffee continues to be an important commodity in the global market, with a significant impact on the economies of producing countries and the daily routines of consumers around the world.In this article, we propose a fresh method of coffee price prediction using the The BLSTM (bidirectional long short-term memory) and CNN (convolutional neural networks) models.We start by collecting historical coffee price data from publicly available sources and preprocess it using feature engineering techniques. The The collected data was then split into training and validation sets and testing sets and feed it into the proposed CNN-BLSTM model.The CNN extraction by using layers the relevant features from the input data and reduce its dimensionality, while the BLSTM layers learn temporal dependencies in the data and capture long-term patterns. The outputs from the BLSTM layers are then fed into fully connected layers, which output the final price prediction.We Use measures like MSE, RMSE, and MAE to measure how far off you are from your target assess how well our suggested model performs (MAE)in both the test and validation data. Our According to the obtained CNN-BLSTM model outperforms several other state-of-the-art machine learning models, including traditional time-series models, on the same dataset.Overall, our approach demonstrates the effectiveness of combining CNN and BLSTM models for coffee price prediction and can be extended to other related forecasting problems.