Optimizing forecasting strategies: Evaluating Holt-Winters models for hotel reservation trends

Abstract

This study undertakes the task of forecasting seasonal time series data, employing Holt-Winters' multiplicative and additive forecasting models. The dataset under scrutiny comprises historical records detailing the daily count of reservations alongside corresponding prices across various room categories in a Vlora-based hotel in Albania, spanning the years 2021 to 2023, post the COVID-19 pandemic. The data originates from the hotel's internal records. Through the utilization of Holt-Winters exponential smoothing techniques, this study discerns distinct trend and seasonal patterns within the daily reservation counts for each room category during the aforementioned period. The process involves establishing initial values and smoothing parameters, crucial for unveiling these patterns. The primary aim is to identify the most effective forecasting method for both the reservation counts across room categories and the price fluctuations. Additionally, an analysis is conducted to compare the influx of foreign citizens arriving in Albania with the consequent impact on increased reservations and pricing within the hotel structure. The central focus of this study is to ascertain the optimal approach while determining the superior methodology for handling such forecasts. Through this comparative analysis, the research seeks to delineate the most favorable approach amidst varied methodologies.