Cash Valuation of Black Tea in the Nuwara Eliya District based on Sensory Quality Attributes: A Case Study

Abstract

The cash valuation of tea; estimated price of tea, decided prior to the auctions, is influenced by sensory assessments from tea brokers and tasters up to a certain degree, although market conditions and customer preferences decide the auction price. This study aims to predict the cash valuation of black tea using sensory quality measures and to identify the key factors that impact these valuations overall and grade specific. This information is crucial for stakeholders to maintain the quality standards of Sri Lankan Tea, supporting the export economy. While past research mainly focused on auction price prediction, few studies have modeled estimated prices using sensory quality parameters. However, using categorical sensory measures is significant in the current study as it was not previously implemented by past researchers. The study analyzed 1,119 tea samples with 13 attributes, finding t hat t ea g rade a nd o rdinal s ensory attributes are important for cash valuation. Due to the attributes being ordinal, numerical encoding was used to predict prices for the entire dataset and for each grade, using statistical and machine learning regression methods. With advanced analysis showing gradient boosting regression as the best predictive model for overall cash valuation of tea, the model achieved a minimum RMSE of 79.75. The study identified t hat t ea g rade, average weight of a tea sample and dry leaf color are essential in cash valuation, with DUST1 and BOPF being the most expensive grades. Cash valuations for these tea grades were observed to be higher when the dry tea leaves were in shades of black.