Applying conditional probability

Advanced Probability: Uncertainty in Data

Maarten Van den Broeck

Senior Content Developer at DataCamp

Understanding conditional probability

• Probability of event B occurring given that event A already happened
• Example:
  • Customer viewed product page (event A)

Understanding conditional probability

• Probability of event B occurring given that event A already happened
• Example:
  • Customer viewed product page (event A)
  • Will customer make purchase? (event B)

Real-world applications of conditional probability

• Customer retention analysis:
  • Will a customer renew given their use of premium features?
• Fraud detection:
  • Is a transaction fraudulent given unusual spending patterns?
• Supply chain optimization:
  • Chance of delay given bad weather?

Revising probabilities with Bayes' theorem

• Revise probabilities in light of new evidence
• Start with: prior probabilities + new data + likelihood

Revising probabilities with Bayes' theorem

• Revise probabilities in light of new evidence
• Start with: prior probabilities + new data + likelihood
• End with: posterior probabilities

Bayesian modeling and decision-making

• Continuous updates as new data comes available
• Manage uncertainty and incomplete data
• Adapt and improve decision-making

Example: fraud detection

• Bayesian model to track credit card fraud
  • Prior probability based on industry data: 2%
• New transaction
  • Sudden spending change behavior
    • Present in 85% of fraudulent transactions
    • Only in 5% of legitimate transactions
  • Updated posterior probability rises significantly, to 25.8%
• Transaction is flagged for review

Let's practice!

Advanced Probability: Uncertainty in Data