A MACHINE LEARNING APPROACH TO PREDICT BLACK FRIDAY SALES

Abstract

In our study of Black Friday sales prediction, we use Multilayer Perceptron
(MLP) models to explore the complex dynamics of this important shopping occasion. Using a
large dataset that includes demographics of the consumer base, product categories, promotions,
and temporal trends, we leverage the power of the MLP model to assess and forecast sales
trends. Because MLP models are flexible, we can capture ambiguities and minute differences
in the sales data, giving us a more nuanced understanding of the dynamics of Black Friday
sales. Our model uses advanced machine-learning algorithms built into the MLP architecture
to learn from past patterns. The programme learns from past sales data to identify underlying
trends and patterns, which improves its ability to anticipate future Black Friday events.
Evaluation measuresthat highlight the model's capacity to precisely capture temporal dynamics
and latent patterns within the sales data are mean absolute error (MAE) and root mean squared
error (RMSE), which act as benchmarks to assess the model's performance.In the end, our
research advances our knowledge of Black Friday customer behaviour and industry trends,
giving merchants useful information to improve their tactics and seize sales opportunities. We
give merchants a strong framework to handle the intricacies of Black Friday by utilising MLP
models, empowering them to make wise decisions and maintain an advantage in the cutthroat
retail market.