Common Use Cases that Cross Industries
Most enterprises, independent of industry, have customers, products, equipment, and employees. These are examples of use cases in each of these areas and can likely be applied in your enterprise.
Note:
This document highlights one use case from each of these focus areas.
Description of the illustration use_cases_cross_industry.png
The following image illustrates the techniques that can be applied for the sample use cases.
The following table summarizes one use case from common cross industry areas and applicable OML techniques and algorithms.
| Area | Use Case | Problem Description | Data | Applicable Techniques and Algorithms | Related Resources |
|---|---|---|---|---|---|
|
Customers |
Customer Lifetime Value |
Calculating the customer lifetime value (LTV) helps one determine the potential earnings for a company from a customer over their complete buying lifetime. It enables businesses to concentrate on long-term bonds with consumers rather than only short-term gains. Businesses anticipate future client expenditure and then modify that figure to consider time, therefore determining LTV. This enables companies to choose which consumers are most important and how best to allocate their money toward marketing and maintaining those relationships. See Wikipedia Customer Lifetime Value. |
SH schema |
|
Blogs:
OML Notebook Example:
|
|
Products |
Cross-sell |
Cross-selling is a sales approach that involves businesses suggesting extra products or services to their current customers. The goal is to increase sales and improve relationships with those customers. Different businesses have their own definitions of cross-selling, which can be shaped by things such as, size of the company, the industry it operates in, and its financial objectives. The main aim of cross-selling is to either boost how much customers spend or improve their loyalty by providing them with useful and relevant products. Companies can use this strategy by having different teams work on it within the organization or by teaming up with partner organizations to explore co-selling opportunities. Companies need to make sure that the new product or service adds value for customers. Big companies often use cross-selling and up-selling strategies together to increase their revenue. They encourage customers to consider buying more expensive items along with related products. See Wikipedia Cross-Selling. |
SH schema |
Association Rules |
Blogs: |
| Equipment |
Predictive Maintenance |
Manufacturers want to plan equipment maintenance at the right time to save replacement costs, prevent failures, and minimize downtime. They use time, usage, and sensor data including climate and vibration to guide decisions. Predicting failure with enough time to act meaningfully presents a major difficulty. For example, if scheduling service takes a week, a 24-hour failure warning is not useful. |
Predictors typically include equipment details (make, manufacturer, in-service date, last maintenance, location, and so on), equipment-specific sensor data (For example, vibration, temperature, pressure), and environmental data (for example, ambient humidity, temperature, indoor or outdoor). The target definition (what we want to predict) must be correct for the current situation. For example, one such target could be an indication (0/1) variable labeled "failed within 5 to 7 days". |
Classification Typical algorithms include: Support Vector Machines (SVM), Generalized Linear Model (GLM), Naive Bayes, Decision Tree, Neural Network, XGBoost. |
Blogs: |
|
Employees |
Retention |
When employees leave a company, it costs a lot to find and train new ones. It’s hard to know who might quit or why, which can affect how well the company works. This use case aims to predict churn risk and suggest retention strategies. It’s like how businesses try to keep customers by offering them more; here, it’s about keeping employees. The goal is to help companies act early to stop employees from quitting. |
HR schema data with employee records. Typically a machine learning model looks at information about employees, such as, how long they’ve worked, their pay, how well they do their job, and if they’re happy. It also uses details like their age, if they’ve been promoted, how often they’re absent, and what jobs are available outside. Text from exit interviews or surveys can be included too. |
Classification Typical algorithms include: XGBoost, Random Forest, Decision Tree, Neural Network. |
Blogs: |
Parent topic: Use Cases Using Oracle Machine Learning