This project analyzes a dataset from the UCI Machine Learning repository to explore the factors influencing customer acceptance of driving coupons. The dataset was collected via a survey on Amazon Mechanical Turk and describes different driving scenarios, including destination, time, weather, passenger, etc., asking whether the user would accept the coupon.
The goal is to identify attributes that distinguish between customers who accepted coupons and those who did not, utilizing visualizations and probability distributions.
coupons.csv: The dataset used for analysis.prompt-2.ipynb: Jupyter Notebook containing the data analysis and visualizations.
The complete analysis, including code and visualizations, can be viewed in this Jupyter Notebook:
https://github.com/zavera/Coupon-Study/blob/main/prompt.ipynb
A detailed analysis is performed to study the usage for two types of coupons - Bar and Restaurants<20 (cheap restaurants). The study shows that different factors affect utility of each type of coupon. Bar coupons are used by drivers that have prior habit of visiting bar frequently in the last month. However the driver is likely to use the coupon if the passenger is not a kid . On the other hand drivers likely to use Restaurant<20 type coupon are motivated by standard meal times ,weather conditions and destination of a non-urgent type . This pattern implies that different coupon type has maximum usability under very different conditions.This data can be leveraged by restaurants to expand their client base.
Overlapping attributes can be analyzed by studying acceptance rate of a different type of coupon. For eg, acceptance rate of a cheap restaurant coupon can be determined for subset of drivers who are likely to also use a bar coupon. This data can be useful for businesses that can be classified as both a bar and a cheap restaurant.
Ambreen Zaver