| ABSTRACT |
Predicting food orders accurately is a perpetual challenge in the restaurant business. This work will explore providing accurate food order prediction through Machine Learning classifications. An accurate forecasting of food orders is key to achieving less food waste, a proactive inventory forecast and happier clients. Increased population and digitalization have significantly increased food demand. Restaurants, University canteen services, catering services
around the world have one conundrum, predicting how much of a food order will come in for any given day. If they fail to show accuracy in predicting food orders, they risk running short of food or excess food waste and costs associated with wasted food and employee management and business profitability. The work utilized a dataset of 5000 entries obtained with variable parameters, such as item name, previous year record, weather, time slot, delivery mode, discount status, holiday status, day of the week, status of the week end and temperature in degree. This work concluded based on
available features, LightGBM appears to a very strong model for predicting food orders. The successful implementation into web application as demonstrated a work able to real-world restaurants. |
