Applying Logistic Regression to Establish a Good Pricing Model for Mobile Phone Manufacturers in the Current Market Landscape using Technical Specifications and User Preference.

Traditionally, and rightfully so, consumers have been forced to part with a premium to own a mobile phone with top-of-the-line features and specifications. Some smartphone manufacturers in 2020 still charge upwards of KES 100,000 for a mobile phone that has a large screen, good battery, fast processor and sufficient storage capacity. However, according to a December article on Android Central, mobile phones with great features are getting significantly affordable. (Johnson, 2020)

A phone’s specifications is a logical way of determining which class it falls under, with the emergence of cheaper manufacturing techniques and parts however, phone pricing models have become more blurry and it is possible for consumers to purchase more powerful smartphones at cheaper prices.

This study intends to explore this hypothesis and predict the relationship between these features and the price of a mobile phone in the current landscape using phone specification, product rating and prices data scraped from a Kenyan e-commerce site.

A supervised learning approach would be useful for this experiment since the data being explored has price labels and categories. Logistic regression is used to classify data by considering outcome variables on extreme ends and consequently forms a line to distinguish them.

The data was collected through scrapping the priceinkenya.com website. For more details on the data acquisition script used please check out Vincent Njonge's Repo https://github.com/lyraxvincent/phones-priceinkenya