Probability & Distributions: Rules, Types & Tourism Applications

TAKEAWAYS:

• Probability rules (addition, multiplication, complement) solve real-world "either/or" and "both" scenarios—like calculating the chance a tourist bus or a flight delay disrupts a travel itinerary.
• Binomial distribution models fixed-trial events (e.g., "How many of 100 hotel bookings will cancel?"), while Poisson handles rare events (e.g., "How many Ncell network outages in a month?").
• The normal distribution (bell curve) dominates tourism data—from guest arrival times to revenue per season—because most real-world metrics cluster around a mean.
• Conditional probability answers "What’s the chance of a delay given heavy rain?"—critical for contingency planning in travel logistics.
• Expected value = (probability × outcome) sums up risk vs. reward (e.g., "Should a trekking agency offer insurance for 50% of clients?").
• Probability distributions turn raw data into actionable insights—like predicting how many Pathao drivers will be available during peak festival traffic.

---

1. Probability Basics: Definitions and Rules

Probability quantifies uncertainty. For tourism, it answers:

• What’s the chance a flight from Kathmandu to Delhi is delayed? (→ affects itinerary planning)
• How likely is a guest to book a return trip? (→ revenue forecasting)

Key Definitions

• Experiment: Any activity with uncertain outcomes (e.g., checking if a guest cancels a booking).
• Sample Space (S): All possible outcomes (e.g., {Booked, Cancelled, No-show}).
• Event (E): A subset of outcomes (e.g., E = {Cancelled}).
• Probability of E: $P(E)=\frac{\text{Number of favorable outcomes}}{\text{Total outcomes}}$.

Probability Rules

Worked Example 1: Ncell Network Reliability Ncell claims 98% uptime. What’s the probability a tourist’s data fails for >30 minutes in a day?

• $P(\text{Failure})=1-P(\text{Uptime})=1-0.98=0.02$ (2%).
• For two consecutive days: $P(\text{Fail both days})=0.02\times 0.02=0.0004$ (0.04%).

---

2. Conditional Probability and Independent Events

Conditional probability asks: "What’s the chance of Event B, given Event A already happened?" Formula: $P(B|A)=\frac{P(A\cap B)}{P(A)}$

Independent vs. Dependent Events

Worked Example 2: Daraz Delivery Delays Daraz’s delivery is delayed 15% of the time. If it rains (30% chance), delays jump to 40%. What’s the probability it rains and the delivery is delayed?

• $P(\text{Rain}\cap \text{Delay})=P(\text{Rain})\times P(\text{Delay}|\text{Rain})=0.30\times 0.40=0.12$ (12%).

---

3. Probability Distributions

Distributions describe how probabilities are spread over outcomes. Three key types for tourism:

A. Binomial Distribution

When to use: Fixed number of trials (n), two outcomes (success/failure), constant probability (p). Formula: $P(X=k)=\left(\genfrac{}{}{0ex}{}{n}{k}\right)p^k(1-p{)}^{n-k}$ Tourism Example: Number of guests who book a package out of 50.

Worked Example 3: Hotel Booking Cancellations A 5-star hotel in Pokhara knows 10% of bookings cancel. For 20 bookings, what’s the probability exactly 3 cancel?

• $n=20$, $k=3$, $p=0.10$.
• $P(X=3)=\left(\genfrac{}{}{0ex}{}{20}{3}\right)(0.10{)}^3(0.90{)}^{17}=1140\times 0.001\times 0.14=0.161$ (16.1%).

---

B. Poisson Distribution

When to use: Rare events over time/space (e.g., complaints, accidents). Formula: $P(X=k)=\frac{e^{-\lambda }{\lambda }^k}{k!}$ Tourism Example: Number of guest complaints per day at a resort.

Worked Example 4: NTC Bus Delays NTC buses average 2 delays per week at a checkpoint. What’s the probability of 4 delays next week?

• $\lambda =2$, $k=4$.
• $P(X=4)=\frac{e^{-2}{2}^4}{4!}=\frac{16\times 0.135}{24}=0.09$ (9%).

---

C. Normal Distribution

When to use: Symmetric, bell-shaped data (e.g., guest arrival times, revenue per season). Formula: $f(x)=\frac{1}{\sigma \sqrt{2\pi }}{e}^{-\frac{(x-\mu {)}^2}{2{\sigma }^2}}$ Key Properties:

• 68% of data within $\mu \pm \sigma$.
• 95% within $\mu \pm 2\sigma$.

Worked Example 5: Tourist Arrivals at Tribhuvan International Airport Daily arrivals average $\mu =500$ tourists with $\sigma =50$. What’s the probability of >600 arrivals tomorrow?

• Convert to Z-score: $Z=\frac{600-500}{50}=2$.
• $P(X>600)=1-P(Z<2)=1-0.9772=0.0228$ (2.28%).

---

4. Expected Value and Variance

• Expected Value (E): Average outcome if an experiment repeats infinitely. $E(X)=\sum x\cdot P(x)$.
• Variance: Measures spread. $\text{Var}(X)=E(X^2)-[E(X){]}^2$.

Worked Example 6: Trekking Agency Profit An agency offers a $100 trekking package with:

• 70% chance of profit ($50),
• 20% chance of break-even ($0),
• 10% chance of loss ($30). Expected Profit: $E(X)=0.70\times 50+0.20\times 0+0.10\times (-30)=35-3=\$32$.

---

In the Real World

1. Ncell Network Reliability

   • Idea Used: Binomial distribution + conditional probability.
   • How: Ncell calculates the probability of network outages during festivals (e.g., Dashain) using historical data. If $P(\text{Outage})=5\%$ normally but jumps to $15\%$ during peak usage, they adjust server loads. Conditional probability: $P(\text{Outage}|\text{Festival})=15\%$.

2. Daraz Delivery Delays

   • Idea Used: Poisson distribution + expected value.
   • How: Daraz models delivery delays as a Poisson process ($\lambda =2$ delays/hour during sales). The expected number of delays in 5 hours is $E(X)=\lambda \times t=2\times 5=10$. This helps them allocate extra drivers during peak hours.

3. Nepal Tourism Board’s Revenue Forecasting

   • Idea Used: Normal distribution + percentiles.
   • How: Monthly tourist arrivals follow a normal distribution ($\mu =120,000$, $\sigma =15,000$). To plan for the "worst 5% of months," they calculate the 5th percentile: $Z=-1.645$, $X=\mu +Z\sigma =120,000-1.645\times 15,000=95,375$ tourists. This ensures they have enough staff for low-traffic months.

---

Exam Tip

1. Spot the Distribution:

   • Fixed trials? → Binomial.
   • Rare events? → Poisson.
   • Bell curve? → Normal (use Z-tables).

2. Watch for "Given" in Questions: Conditional probability questions often use phrases like "if," "given," or "knowing that." Always rewrite $P(B|A)$ as $\frac{P(A\cap B)}{P(A)}$.

3. Real-World Twist: Exams love tourism scenarios. If a question mentions "hotel bookings," "flight delays," or "guest complaints," think:

   • Binomial for counts (e.g., cancellations).
   • Poisson for rare events (e.g., complaints).
   • Normal for symmetric data (e.g., arrival times).

4. Expected Value Shortcut: For discrete data, list all outcomes and multiply by their probabilities. For continuous data (normal), use the mean $\mu$.

5. Avoid Common Mistakes:

   • Addition vs. Multiplication: $P(A\cup B)\ne P(A)\times P(B)$.
   • Independent Events: Only multiply if events are independent!
   • Z-Table Errors: Remember $P(X>a)=1-P(X<a)$.

---

Final Visual Summary

In the real world

• Ncell uses the Poisson distribution to predict rare network outages during festivals (e.g., Dashain, Tihar) and adjusts server capacity accordingly. For example, if the average outage rate is λ=2 per week, they calculate the probability of 4+ outages during peak usage (P(X≥4) ≈ 15%) to prepare backup systems.

• Pathao applies conditional probability to estimate ride demand during heavy rain. If P(Rain)=30% and P(Delay|Rain)=40%, they dynamically increase driver incentives in rainy areas to maintain service reliability.

• Nepal Tourism Board relies on the normal distribution to forecast tourist arrivals at Tribhuvan International Airport. Using historical data (μ=500 tourists/day, σ=50), they allocate airport resources (e.g., immigration counters, baggage handlers) based on the 95% confidence interval (μ±2σ = 400–600 tourists/day).