This post offers a brief introduction to three important predictive analytics application areas – customer analytics, operational analytics, and threat and fraud analytics. Further, the four key steps of the predictive analytics process are outlined.
- Three important application areas of predictive analytics
- The predictive analytics process (4 steps)
You may also be interested in What is data mining?
Three important application areas of predictive analytics
This section outlines three important application areas of predictive analytics: customer analytics, operational analytics, and threat and fraud analytics (see Hill, 2013).
1. Customer analytics
Understand your customers and what they are likely to do:
1.1. ACQUIRE customers more efficiently: Understand who are your best customers; connect with them effectively; take actions to maximize sales.
1.2. GROW the value of existing customers through personalized up-sell/cross-sell: Understand the best mix of products/services your customers need; maximize revenue from customers; optimize interaction strategy with customers across digital media platforms.
1.3. RETAIN customers for a long time: Understand what value your customers seek/gain from your products/services; keep your best customers happy (vs. the value/experience they seek); take action to retain customer loyalty (build relationships).
2. Operational analytics
Manufacturing, supply chain, service industry–people, processes, assets:
2.1. PLAN operations: Allocate expenditures effectively; manage inventory; identify the economic and social impact.
2.2. MANAGE daily operations: Understand causes of failure; improve employee performance (productivity); reduce upkeep cost.
2.3. MAXIMIZE operational performance: Improve asset (machine) performance; reduce operational costs; drive operational excellence in procurement, development, availability, distribution.
3. Threat and fraud analytics
Detect and mitigate threat/fraud/suspicious/anomalous transactions:
3.1. MONITOR system/organizational environment to identify fraudulent patterns: Identify data/asset leaks; enforce compliance; leverage insights in critical business functions.
3.2. DETECT suspicious activity/threat: Identify fraudulent patterns; identify collusive merchants and employees; identify unexpected or unusual transactions.
3.3. CONTROL outcomes to deliver the best response to reduce exposure or loss and maximize the impact of remedial actions: Take action in real time to prevent abuse; reduce claim processing time; alert customers about fraudulent transactions.
The predictive analytics process (4 steps)
This section outlines the steps to build a predictive analytics model to predict the propensity of a customer to churn, as an example.
1. DEFINE targeted outcome/dependent variable.
Output data: targeted variable (the likelihood to churn).
2. CAPTURE/collect customer data required to train the predictive model.
Input data: predictors or independent variables.
Demographics (descriptive) data: age, gender, recent activity, satisfaction, marital status.
Transactional (behavioral) data: # orders.
3. PREDICT outcome/output.
Split the data set into training set (~80%) and testing set (~20%).
Training set: apply an algorithm to the training data to construct a set of rules that predict the value of the churn (target) attribute (predicted value or dependent variable).
Testing set: deploy model (scoring).
Assess the accuracy of the model (ratio of correct predictions to actual outcomes).
4. ACT on insights/predictions.
Key references
Hill, Brad. [bradhill14]. (2013, Jul 9). What is Predictive Analytics? [Video]. YouTube. https://www.youtube.com/watch?v=0KXME_y-3QA
Related content
Decision tree for classification problems
Google Data Analytics Professional Certificate quiz answers
Google IT Support Professional Certificate quiz answers
How to break into information security
IT career paths – everything you need to know
The Security Operations Center (SOC) career path
What is the Google Data Analytics certification?
Back to DTI Courses
Other content
1st Annual University of Ottawa Supervisor Bullying ESG Business Risk Assessment Briefing
Disgraced uOttawa President Jacques Frémont ignores bullying problem
How to end supervisor bullying at uOttawa
PhD in DTI uOttawa program review
Rocci Luppicini – Supervisor bullying at uOttawa case updates
The case for policy reform: Tyranny
The trouble with uOttawa Prof. A. Vellino
The ugly truth about uOttawa Prof. Liam Peyton
uOttawa engineering supervisor bullying scandal
uOttawa President Jacques Frémont ignores university bullying problem
uOttawa Prof. Liam Peyton denies academic support to postdoc
Updated uOttawa policies and regulations: A power grab
What you must know about uOttawa Prof. Rocci Luppicini
Why a PhD from uOttawa may not be worth the paper it’s printed on
Why uOttawa Prof. Andre Vellino refused academic support to postdoc