I am working on a product matching model.
- A store has many products like creams, perfumes, other beauty products.
- Based on product properties I have to create bundles of it so we can sell more product at once while giving a small discount to the customer.
- I can not make a project based on Collaborative filtering because the goal is to have similar styles matched together ex.: 1.) Hugo Boss perfumes and creams, 2.) Summer lipstick collection (any brand). That is what is our customer base is all about that we have big sales and nice discounts. Currently, these products are matched manually.
#1. Product Matching [Clustering Algorithm, Unsupervised Learning]
- QUESTION What Unsupervised Learning algorithm you recommend if I don't know how many products groups there should be. We have currently 4000 products. (I am currently using: Agglomerative Hierarchical Clustering). The goal would be to get an ALL PRODUCT x ALL PRODUCT matrix with 1 number in each matrix cell that defines the closeness of the 2 compared product.
#2. Salles Representative want to define some parameters of the products (price, product types)
- My goal if they pick 1 product (they copy the product_id to the cell) -> that an excel sheet looks up for them that
- They want to define things like: product type(), Quantity, Price, Margin %, Review Score(at least 4/5 star on our site), Wishes(it is on people's Wishlist) -> ex.: the sales person can filter it that he only wants lipsticks in an excel sheet -> so he gets the closest matching lipsticks
#3. Supervised Machine Learning Regression
- Predicts sales based on previous product bundles that the sales representative created even before this algorithm.