• Carolina Duque Chopitea

Using Machine Learning to Identify Cross-selling Opportunities at Apprentice Chef | Machine Learning


Python code here


Company Background & Analysis Objective Apprentice Chef, Inc. is a home-prep food delivery company, developed for the busy professional that has little to no skills in the kitchen, they offer a wide selection of daily-prepared gourmet meals delivered directly to your door.


In an effort to diversify their revenue stream, Apprentice Chef, Inc. has launched Halfway There, a cross-selling promotion where subscribers receive a half bottle of wine from a local California vineyard every Wednesday (halfway through the work week). As such the purpose of this study is to predict which customers will subscribe to "Halfway There". Halfway There has been offer to all the customers in the datasets, yet not all have taken the promotion (variable CROSS_SELL_SUCCESS).

The goal is therefore (1) to analyze the sample data, (2) develop insights and (3) build and compare machine learning models, and determine the best model to predict cross-sell success.

Some of the questions this study aims to answer include:

  • Is there a relationship between customers who sign up for Halfway There and the total meals they purchase or revenue?

  • Are those customers enrolled in weekly plans more likely to cross-sell?

  • What are the common characteristics among subscribers? What features drive cross-sell success?



Insight 1:


People either personal or professional emails are more likely to subscribe to Halfway There than those with junk email. This may mean that people with junk emails are perhaps not as engage with the platform as other users. For example, when the promotion for the cross selling begun, they may not even find out about it because they didn’t receive it in their primary email.


Insight 2:


From the model (pruned tree model) it can be observed that following recommendations is the most important variable to determining cross selling. When a customer follows recommendation 35% or more of the time, they are more likely to subscribe.


Actionable Recommendation:


Further information about the customers will be essential to understand where the potential cross selling opportunities for Halfway There are. Gathering data like age and gender, and other demographics may help improving our prediction since demographics can play an important role in wine purchasing decision.


• Bay Area residents who visit Wine Country are married, Caucasian, and affluent adults between the ages of 35-54 (Lottridge, 2018). This segment represents a potential cross selling opportunity.


• Wine sales are expected to increase in the USA, being women the major consumers of this category (57%) (Galante, 2019).


• Location information is important since Households in the Bay Area spend more than $700 million on wine annually. The counties ranking the highest by spend are Santa Clara ($177.8M), Alameda ($139.8M), San Francisco ($96.6M), and Contra Costa ($94.2M) (Lottridge, 2018). Effort to promote Halfway can be centered on these cities.


• In addition, millennials are emerging as the primary wine consumer and are more likely to consume it at home. Today, 80% of all wine consumption is at home.

This information should not be too costly to gather and can be asked when the person signs up for Apprentice Chef.





Classification and Regression Tree: best model to predicting cross-sell success, with an AUC Score of 0.726



References



Conway, J. (2019, May 11). Customer share of Blue Apron as of 2018, by age. Statista, pp. From:https://www.statista.com/statistics/948016/blue-apron-customer-share-by-age-worldwide/.

Galante, M. (2019, April 2). 10 Wine Trends to Watch for in 2019. Dimentional Inisghts, pp. From:https://www.dimins.com/blog/2019/04/02/10-wine-sales-trends-watch-2019/.

Gross, D. (2010, November 16). What your e-mail address says about you. CNN, p. From:https://www.cnn.com/2010/TECH/web/11/16/email.users/index.html.

Lottridge, S. ( 2018, May 5). GET TO KNOW THE BAY AREA WINE CONSUMER. Hearst Bay Area, pp. From:https://marketing.sfgate.com/blog/get-to-know-the-bay-area-wine-consumer.

ProtonMail. (n.d.). About ProtonMail. pp. From:https://protonmail.com/support/knowledge-base/business/.

Carolina Duque Chopitea