Chain Rule Product Rule

Deep Breakdown

What it tells me (Situation Awareness): The chain rule decomposes the joint probability of two events into a product of a marginal and a conditional. It generalizes to $n$ events: $P(A_1 \cap \ldots \cap A_n) = P(A_1) \cdot P(A_2 \mid A_1) \cdots P(A_n \mid A_1, \ldots, A_{n-1})$. Fundamental to Bayesian networks, language models, and Hidden Markov Models.

Flowchart

Real-world use

Where it is used

Bayesian inference: computing likelihoods. Language models: $P(w_1, w_2, \ldots, w_n) = \prod P(w_i \mid w_1, \ldots, w_{i-1})$. Naive Bayes: factoring $P(\text{features} \mid \text{class})$.

Problem it solves

How do I compute or factorize the joint probability of multiple events without knowing the full joint distribution?

How it solves the problem

Rewrites the joint as a product of simpler, estimable terms — a marginal and a conditional — which are often easier to compute or estimate from data.

Example in a business

Customer journey: $P(\text{purchase} \cap \text{cart}) = P(\text{purchase} \mid \text{cart}) \times P(\text{cart})$. Used to decompose funnel conversion rates into conditional step probabilities.

The formula, piece by piece

$P(A \cap B) = P(A \mid B) \cdot P(B)$

• $P(A \cap B)$: joint probability of both events.
• $P(A \mid B)$: conditional probability (how likely $A$ is given $B$).
• $P(B)$: marginal probability of $B$.
• Equivalent form: $= P(B \mid A) \cdot P(A)$ (symmetric).

The formula in words

"The joint probability of A and B equals the probability of B times the probability of A given B."

Critical Concepts

• This is exact, not an approximation (unlike Naive Bayes which assumes independence).
• Generalizes: $P(A,B,C) = P(C \mid A,B) P(B \mid A) P(A)$.
• Basis of Bayes' theorem: $P(A \mid B) = P(B \mid A) P(A) / P(B)$.
• Rearranging: $P(A \mid B) = P(A \cap B) / P(B)$ — back to conditional probability definition.

Step-by-step calculation (tiny example)

Step 1: Known rates P(add to cart) = 0.30. P(purchase | cart) = 0.40.

Step 2: Chain rule $P(\text{cart} \cap \text{purchase}) = P(\text{purchase} \mid \text{cart}) \times P(\text{cart})$ $= 0.40 \times 0.30 = 0.12$

Step 3: Interpret 12% of all visitors both add to cart AND purchase.

Code examples

Beginner example (Python)

# Funnel decomposition via chain rule
p_cart = 0.30   # P(add to cart)
p_purchase_given_cart = 0.40   # P(purchase | cart)
p_purchase_and_cart = p_purchase_given_cart * p_cart
print(f"P(cart ∩ purchase) = {p_purchase_and_cart:.2f}")   # 0.12

Beginner example (R)

p_cart <- 0.30
p_purchase_given_cart <- 0.40
p_purchase_and_cart <- p_purchase_given_cart * p_cart
cat("P(cart ∩ purchase) =", p_purchase_and_cart, "\n")  # 0.12

Professional example (Python)

import numpy as np
from typing import Sequence

def chain_rule_decomposition( conditional_probs: Sequence[float], marginal_first: float ) -> float:
    """Compute joint probability via chain rule for sequential events. P(A1 ∩ ... ∩ An) = P(A1) * P(A2|A1) * P(A3|A1,A2) * ... Args: conditional_probs: P(Ak | A1,...,A_{k-1}) for k=2,...,n. marginal_first: P(A1), probability of first event. Returns: Joint probability of all events in sequence. """
    joint = marginal_first
    for cond_prob in conditional_probs:
        joint *= cond_prob
    return joint   # E-commerce funnel analysis
funnel_steps = [ ("Session start", 1.00), ("P(browse | start)", 0.75), ("P(add_to_cart | browse)", 0.30), ("P(checkout | cart)", 0.55), ("P(purchase | checkout)", 0.80),
]
cumulative = 1.0
print("Funnel cumulative probabilities:")
for step_name, step_rate in funnel_steps:
    cumulative *= step_rate
print(f" {step_name:<35} {cumulative:.4f} ({cumulative*100:.2f}%)")

Source: AIO p. 84

#probability-statistics#probability-theory#conditional-probability-and-independence