Brand_positioning

• 

  Position of a lipstick brand relative to its competition

• 

  Table of keyphrase counts of each brand

Inspiration

Having the Amazon customer reviews, I build a recommender system that is useful for both of the customer and company. The problem is to investigate customer reviews with NLP tools, and derive some insight. For one thing, the problem aims to reduce the customer risks. For another thing, the problem is for a company to increase its product efficacy and quality to outgo other competitors in the market.

What it does

Identify gaps in the market

Identify potential partners

Clarify perceptual problems of a brand

How I built it

I pulled out the Keywords from the reviews using Rake (A module in Python). Then Used SVD to reduce the dimensionality of the data matrix to a 2D subspace. The resulting 2D graph has some brands and their relative attributes as specific points on the plane.

Challenges I ran into

Preprocessing,

deriving the key phrases,

creating appropriate input to feed into the brand positioning algorithm.

Accomplishments that I'm proud of

Visualizations of the data such that either customer or company can benefit from that because it has some insights for both of them. A customer can benefit it by observing how some brands are close to specific attributes. A company can benefit it by observing how their brand is far from some attributes and compete with other brands.

What I learned

I learnt not to rely on only one approach and at least try a couple of methods.

What's next for Brand_positioning

An improvement could be to replace Rake with another method that takes into account similarities of the words. (Something like MAUI in Java, which at present is not out there in Python but it definitely worth working on!)

Built With

• particle
• python