Though some of these are actually optimization…
This came out of at Fast.ai ppt that can be found here. Granted they only provided the list. You will notice an ethics deck they have uploaded as well, I encourage you to review at it. I have a few ethics slides in my data science talk, but the fast.ai gang hit way harder than I typically do. I admit, shock is a good way to wake people into thinking about what they are doing.
In their ML Applications deck they have a list of applications by industry, below I have them listed out and what I hope is to present either an elevator pitch of what each one is, or and executive overview of each and links to more info. This post will be made of a lot of quotes and references, that is kind of the point, no original content will come from me as I am not the creator of much data science, just a user of it, though i am sure i will add commentary. This will be a series of blogs posts, and clearly each post has the potential for being very long even with just a brief summary and a few links.
The funny thing about solving a data science problem is that their are many ways to solve it, so i don’t expect this to be 100% comprehensive, i try to find what appears to be a canonical solution, though that does not mean you cannot stuff everything into a neural net and close your eyes, which is what everyone appears to be doing these days…
Enjoy
Horizontal Applications
Horizontal Applications
“A horizontal application is any software application that targets a large number of users with different knowledge and skill sets. Because these types of applications can extend across markets and be used in a range of industries, they typically do not offer market-specific features.“[1a]
Marketing Applications
Predictive Lifetime Value LTV/CLV
“The goal of predictive CLV is to model the purchasing behavior of customers in order to infer what their future actions will be.” [1]
“Historical CLV has several drawbacks, the most important of which being that, since it is the sum of past revenue or profit for a particular customer or group, it only provides insight into what has already occurred, and, thus, sheds little insight into the value of new subscribers. Predictive CLV, however, with its ability to incorporate expected retention, allows marketers to obtain several key insights, including what types of subscribers will be the most profitable over a specific time period” [2]
Potential Models:
More Learning:
https://github.com/datascienceinc/pydata-seattle-2017/tree/master/lifetime-value
Implementing and Training Predictive Customer Lifetime Value Models in Python
“Share of wallet (SOW) is a marketing metric used to calculate the percentage of a customer’s spending for a type of product or service that goes to a particular company. For example, if a customer spends $60 a month at fast food restaurants, and $30 of that amount is spent at McDonald’s, McDonald’s has a 50% SOW for that customer. The term is sometimes expressed as wallet share.”[4] “Using the Wallet Allocation Rule, the grocer calculates its average share of wallet and that of its three main competitors.” [6]
Definition of the Wallet [4]
Potential Models:
Assessing Size and Share of Customer Wallet
Duke University, Size and Share of Customer Wallet
Churn
“Customer attrition, also known as customer churn, customer turnover, or customer defection, is the loss of clients or customers.” … [9]
“Companies usually make a distinction between voluntary churn and involuntary churn. Voluntary churn occurs due to a decision by the customer to switch to another company or service provider, involuntary churn occurs due to circumstances such as a customer’s relocation to a long-term care facility, death, or the relocation to a distant location. In most applications, involuntary reasons for churn are excluded from the analytical models. “[9]
“in games where players must be engaged on a day-to-day basis, a player who doesn’t login within 24 hours may be considered a churner.”[12]
Potential Models:
Customer Segmentation
“Customer segmentation is the process of dividing customers into groups based on common characteristics so companies can market to each group effectively and appropriately. … Segmentation allows marketers to better tailor their marketing efforts to various audience subsets.” [13]
“Create and communicate targeted marketing messages that will resonate with specific groups of customers, but not with others (who will receive messages tailored to their needs and interests, instead). Select the best communication channel for the segment, which might be email, social media posts, radio advertising, or another approach, depending on the segment.” [13]
Potential Models:
Customer Segmentation in Python – PyConSG 2016
Product Mix
(Optimization)
“The Product Mix also called as Product Assortment, refers to the complete range of products that is offered for sale by the company. In other words, the number of product lines that a company has for its customers is called as product mix.”[16]
“The product mix could include several lines of products or individual products that do not fall into a line. For example, if a company owns a line of hygiene products and also owns a line of house cleaning products, all of those products together would constitute the product mix for the company. Each line would combine with the other to come up with the total mix.” [17]
Potential Models:
Cross Selling
“Cross-selling is a sales technique used to get a customer to spend more by purchasing a product that’s related to what’s being bought already. It’s easy to confuse cross-selling with upselling. Cross-selling involves offering the customer a related product or service, while upselling typically involves trading up to a better version of what’s being purchased.
Amazon reportedly attributes as much as 35 percent of its sales to cross-selling through its “customers who bought this item also bought” and “frequently bought together” options on every product page. That approach allows a retailer to prompt a shopper to buy a compatible – or necessary – product.”[20]
Potential Models:
Recommendation Algorithms
“A recommender system or a recommendation system (sometimes replacing “system” with a synonym such as platform or engine) is a subclass of information filtering system that seeks to predict the “rating” or “preference” a user would give to an item.
Recommender systems typically produce a list of recommendations in one of two ways – through collaborative filtering or through content-based filtering (also known as the personality-based approach). Collaborative filtering approaches build a model from a user’s past behaviour (items previously purchased or selected and/or numerical ratings given to those items) as well as similar decisions made by other users. Content-based filtering approaches utilize a series of discrete characteristics of an item in order to recommend additional items with similar properties” [23]
Potential Models:
K Nearest Neighbor[24]
Deep Neural Net[24]
Collaborative based(python)[25]
Content Based(python)[25]
Building a Recommender System in Azure Machine Learning Studio
Up-Selling
“Do you want Fry’s with that? “[26] You have just experienced an up-sell!
“Upselling is a sales technique where a seller induces the customer to purchase more expensive items, upgrades or other add-ons in an attempt to make a more profitable sale. While it usually involves marketing more profitable services or products, it can be simply exposing the customer to other options that were perhaps not considered”[29]
“Using data correlation techniques on its massive database of over 150 million customers, Amazon is able to gain insights on past purchases, reviews, customer preferences and product popularity to make relevant and personalised recommendations to users that match their buying history and preferences.”[26]
Potential Models:
Channel Optimization
The cusp of this is finding the correct marketing channel to reach your consumers, e.g. social media, email, news print, commercials, sky writing.
“determine the optimal (usually meaning most profitable) allocation of marketing resources across available channels” [29]
Potential Models:
Discount Targeting
“Target Marketing involves breaking a market into segments and then concentrating your marketing efforts on one or a few key segments consisting of the customers whose needs and desires most closely match your product or service offerings. ” [32] and presumably hitting these customers with discounts such as coupons.
Potential Models:
Reactivation Likelihood
Lost customer due to churn returning. This is related to customer Churn.
Potential Models:
Ad-words Optimization and Buying
This is similar in concept to channel optimization in that you are looking for the impact of a word or series of words. Most methods of customer contact will perform this analysis for you so little will be done from the user side. But, basic statistics could be applied, ANOVA, A/B test etc.
Potential Models: