Forecast - Planning - Recommendations - Paper Reads

Paper #1 - Maximizing Store Revenues using Tabu Search for Floor Space Optimization

Key Notes

• Floor space is a valuable and scarce asset for retailers
• Connected multi-choice knapsack problem with an additional global
• constraint and propose a tabu search based metaheuristic that exploits the
• multiple special neighborhood structures
• Over the last decade, the number of products competing for limited space increased by up to 30%
• The product mix of categories, merchandising rules, sales patterns and characteristics of display furniture 
• (1) develop a statistical model to measure the space elasticity; and 
• (2) formulate and solve an optimization problem for each store to determine the optimal assignment of planograms to maximize total revenue subject to certain business constraints

Paer #2 - Reversing ShopView analysis for planogram creation

Key Notes

• ShopView can build the planogram without the need of manually creating it in software
• OCR in the identification of products
• Planograms specifies the absolute physical locations of the products, and the amount of space each type of product should occupy
• Planogram compliance using template images
• Vision - Object Recognition based on attributes, Template and Feature Matching, Optical Character Recognition (OCR)
• Custom Dictionary - Implementing a custom dictionary for the OCR engine seemed a good strategy since at first glance it would improve the performance of the OCR algorithm

Paper #3 - Deep Learning based Recommender System: A Survey and New Perspectives

Key Notes

• Collaborative €ltering makes recommendations by
• learning from user-item historical interactions, either explicit (e.g. user’s previous ratings) or implicit feedback (e.g. browsing history)
• Content-based recommendation is based primarily on comparisons across items’ and users
• Hybrid model refers to recommender system that integrates two or more types of recommendation strategies
• Strengths of deep learning based recommendation models - Nonlinear Transformation, Sequence Modelling

Paper #4 - Fashion Retail: Forecasting Demand for New Items

Key Notes

• Merchandising Factors - Discount, Visibility, Promotion
• Derived Features - Age of Style, Trend and Seasonality, Cannibalisation

Paper #5 - Time Series Forecasting With Deep Learning: A Survey

More Reads

• A Survey Paper on Recommender Systems
• A Comparative Analysis of RNN and SVM
• Accurate Demand Forecasting for Retails with Deep Neural Networks
• A Worrying Analysis of Probabilistic Time-series Models for Sales Forecasting
• Implementing Reinforcement Learning Algorithms in Retail Supply Chains with OpenAI Gym Toolkit
• Sales Demand Forecast in E-commerce using a Long Short-Term Memory Neural Network Methodology

Keep Exploring!!!