Use of Bayes' theorem to calculate simple conditional probabilities

Welcome to the World of Bayesian Statistics!

Hello there! Today, we are diving into one of the most powerful tools in an actuary’s toolkit: Bayes' Theorem. This chapter is the foundation of the "Bayesian Statistics" section of your CS1 exam. Don't worry if probability has felt a bit abstract in the past—Bayes' Theorem is actually very logical. It’s essentially the mathematics of updating your beliefs when you get new information.

Think of it like being a detective. You start with a "hunch" (your initial probability), you find a new piece of evidence, and then you use that evidence to calculate a new, more accurate probability. Let’s break it down step-by-step!

1. The Building Blocks: What is Conditional Probability?

Before we look at the formula, let’s refresh a concept you’ve likely seen before: Conditional Probability. This is the probability of an event occurring, given that another event has already happened.

We write this as \( P(A|B) \), which we read as "the probability of A given B."

Quick Review Box:
• \( P(A) \): The probability that event A happens (on its own).
• \( P(A \cap B) \): The probability that both A and B happen.
• \( P(A|B) = \frac{P(A \cap B)}{P(B)} \): The formula for conditional probability.

Analogy: Imagine you are looking at the probability that it will rain (\( A \)). If I tell you that the sky is currently covered in dark, heavy clouds (\( B \)), your estimate of rain will probably go up. You are calculating \( P(Rain | Clouds) \).

2. Introducing Bayes' Theorem

Bayes' Theorem allows us to "flip" conditional probabilities. If we know \( P(B|A) \), Bayes' Theorem helps us find \( P(A|B) \).

The Official Formula:
\( P(A|B) = \frac{P(B|A) \times P(A)}{P(B)} \)

To make this easier to remember for the CS1 exam, we give these parts special names:

1. Prior Probability \( P(A) \): What we believed before seeing the evidence.
2. Likelihood \( P(B|A) \): How likely the evidence is, assuming our hypothesis \( A \) is true.
3. Posterior Probability \( P(A|B) \): Our new, updated belief after seeing the evidence.
4. Normalising Constant \( P(B) \): The total probability of the evidence occurring under all possible scenarios.

Did you know? Bayes' Theorem was named after Thomas Bayes, an 18th-century minister. He never actually published it! His friend Richard Price found his notes and published them after Bayes died, realizing how revolutionary the idea was for science and statistics.

3. The Law of Total Probability (The Denominator)

Students often find the bottom part of the fraction, \( P(B) \), the trickiest. To find the total probability of \( B \), we have to consider all the different ways \( B \) could happen.

If there are only two possibilities for \( A \) (let's say \( A \) happens or it doesn't, which we call \( A' \)), then:
\( P(B) = P(B|A)P(A) + P(B|A')P(A') \)

Step-by-Step Breakdown:
Imagine you are an insurance actuary. You have "Good Drivers" (\( G \)) and "Bad Drivers" (\( B \)). You want to know the probability that a randomly selected person has a claim (\( C \)).
1. Find the probability a Good Driver has a claim: \( P(C|G)P(G) \).
2. Find the probability a Bad Driver has a claim: \( P(C|B)P(B) \).
3. Add them together to get the total probability of a claim: \( P(C) \).

4. A Real-World Actuarial Example

Let’s put it all together with a classic exam-style scenario. Don't worry if this seems tricky at first; just follow the steps.

Scenario: An insurance company knows that 10% of its policyholders are "High Risk" and 90% are "Low Risk." The probability of a "High Risk" driver having an accident in a year is 0.4. For a "Low Risk" driver, it is 0.2. A policyholder just had an accident. What is the probability they are "High Risk"?

Step 1: Identify what we know (The Priors and Likelihoods)
\( P(High) = 0.10 \)
\( P(Low) = 0.90 \)
\( P(Accident | High) = 0.40 \)
\( P(Accident | Low) = 0.20 \)

Step 2: Calculate the Total Probability of the evidence (The Denominator)
\( P(Accident) = (0.40 \times 0.10) + (0.20 \times 0.90) \)
\( P(Accident) = 0.04 + 0.18 = 0.22 \)

Step 3: Apply Bayes' Theorem (The Posterior)
\( P(High | Accident) = \frac{P(Accident | High) \times P(High)}{P(Accident)} \)
\( P(High | Accident) = \frac{0.40 \times 0.10}{0.22} = \frac{0.04}{0.22} = 0.1818 \)

Key Takeaway: Before the accident, we thought there was a 10% chance they were high risk. After the accident, that probability jumped to roughly 18.2%. The evidence changed our belief!

5. Common Mistakes to Avoid

Even the brightest students make these mistakes, so keep an eye out for them:

• Confusing \( P(A|B) \) with \( P(B|A) \): Always ask yourself: "What do I already know for sure?" That goes after the vertical bar.
• Forgetting to multiply by the prior: It’s not just about how likely the evidence is; it’s about how likely the original event was in the first place.
• Probabilities not summing to 1: Your initial categories (like "High Risk" and "Low Risk") must account for 100% of the population.

6. Summary and Quick Review

Wait! Before you move on, remember:

• Bayes' Theorem is used to update probabilities when new data is available.
• Prior is the "before" probability; Posterior is the "after" probability.
• Use the Law of Total Probability to find your denominator—it's just the sum of all possible paths to the evidence.
• In CS1, drawing a tree diagram can be a lifesaver! Each branch represents a piece of the calculation, making it much harder to lose track of your numbers.

You've got this! Bayes' Theorem is a logic puzzle. Once you identify your Priors and your Likelihoods, the rest is just simple arithmetic. Practice a few tree diagrams, and you’ll be a Bayesian pro in no time!