DATA-DRIVEN INVENTORY MANAGEMENT IN FASHION RETAIL - A MACHINE LEARNING APPROACH TO DEMAND FORECASTING

Aug 5, 2026

This study develops a machine learning model to enhance inventory management at the “Kilt” clothing store by accurately predicting annual product sales, preventing overstock, and maximizing profitability. Analyzing historical sales data from 2009 to 2023, multiple regression models were evaluated, with Random Forest identified as the most effective. Model forecasts for 2024 estimate profits reaching 2.2 million versus 1.4 million in 2023. The approach supports data-driven inventory decisions and improves overall store profitability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages39

Reading Time1 h 38 min

AccessFree to download

Introduction

Research Questions (RQ1-RQ2)

Literature Review

Problem Description and Data Manipulation

A Work Project, presented as part ofthe requirements for the Award of a Master’s degree in Management from the Nova School of Business and Economics. DATA-DRIVEN INVETORY MANAGEMENT IN FASHION RETAIL: A MACHINE LEARNING APPROACH TO DEMAND FORECASTING TOMMASO GUASTI Work project carried out under the supervision of: Rongjiao Ji 11/06/2024 Abstract This study develops a machine learning model to enhance inventory management at the “Kilt”clothing store by accurately predicting annual product sales, preventing overstock, and maximizing profitability. Analyzing historical sales data from 2009 to 2023, I evaluated multiple regression models, identifying the Random Forest as the most effective. The model forecasts 2024 profits to reach 2.2 million, significantly higher than 2023 ’s 1.4 million. By leveraging past sales data and advanced predictive modeling, the study provides strategic insights to optimize inventory decisions and improve overall store profitability. Keywords Sales Forecasting, Machine Learning, Inventory Management, Profit Optimization, Random Forest Model, Retail Analytics. This work used infrastructure and resources funded by Fundação para a Ciência e a Tecnologia (UID/ECO/00124/2013, UID/ECO/00124/2019 and Social Sciences DataLab, Project 22209), POR Lisboa (LISBOA-01-0145-FEDER-007722 and Social Sciences DataLab, Project 22209) and POR Norte (Social Sciences DataLab, Project 22209) . 1 Introduction To ensure sustainability and profitability in the fast-paced world of fashion retail, effective inventory management is crucial. The sector requires sophisticated stock-level management solutions because of its frequently changing trends and variable consumer preferences. Demand forecasting can be effectively addressed by data-driven decision-making, particularly when machine learning is employed. This paper uses machine learning algorithms to enhance demand forecasting. It focuses on transactional data from the Rome-based clothing store “Kilt”, using sales data and specific product attributes to guide the analysis. Conventional inventory management in small to medium sized fashion retail sometimes ignores the multitude of elements impacting fashion purchase trends in favor of oversimplified statistical methods or intuitive projections based on historical sales. The research holds relevance as it can close this gap by creating a model that can predict the quantity of products sold, for each type of product, with accuracy, over a one-year period, thereby significantly increasing inventory efficiency. This capability addresses a critical point for small stores like “Kilt”, where the owner must place yearly orders at the beginning of the year and always over-orders due to fears of stockouts. Better forecasting models should improve financial performance and decrease overstock. Key questions guide the research: RQ1. What is the best model for predicting annual product sales at the "Kilt" clothing store, thereby maximizing profit? The demand for different products is predicted using machine learning models like Random Forest and K-nearest neighbors (KNN) . These models offer insights that improve inventory management and sales tactics. RQ2. What types and quantities of products should the Kilt clothing store purchase for 2024 based on predictive sales modeling? This issue seeks to maximize efficiency and profitability in inventory procurement over the next year by using the insights from machine learning forecasts to real-world decision making. 2 Literature Review Besides this work, others also engage in examining the use of machine learning models for sales prediction. An overview of the state of the research on machine learning models for sales prediction is given in this section. Ren et al. (2020) demonstrated that the accuracy of demand forecasting is greatly improved when big data analytics are integrated into retail operations, especially for fashionable products. Ashraf (2022) pe

FAQs

What problem does the study address in fashion retail inventory management?

It addresses the need for accurate demand forecasting to support inventory decisions in a fast-changing fashion context, reducing over-ordering and stockout risk.

Which machine learning model performs best in the study?

Random Forest is identified as the most effective model after evaluating multiple regression approaches on historical sales data from 2009 to 2023.

How does the model influence business outcomes for Kilt in 2024?

It forecasts product sales and expected profits to help optimize purchasing quantities, with the study projecting 2024 profits of 2.2 million compared with 1.4 million in 2023.

Aug 5, 2026

Upload by River Wang