What is Statistical Modelling?

S3-SA3-0198

Grade Level:

Class 9

AI/ML, Data Science, Physics, Economics, Cryptography, Computer Science, Engineering

Definition

What is it?

Statistical modelling is like creating a mathematical map or a simplified story to understand patterns and relationships hidden in data. It helps us predict future outcomes or understand why certain things happen based on past information.

Simple Example

Quick Example

Imagine you want to predict how many runs a cricket team will score in the next match. You look at their scores from previous matches, how well their top batsmen performed, and the pitch conditions. Statistical modelling helps you use all this past data to make an educated guess for the upcoming match.

Worked Example

Step-by-Step

Let's say you want to predict your mobile data usage (in GB) based on how many hours you stream videos.

Step 1: Collect data. You note down your video streaming hours and data usage for the last 5 days:
Day 1: 1 hour streaming, 1.5 GB data
Day 2: 2 hours streaming, 2.5 GB data
Day 3: 0.5 hours streaming, 1 GB data
Day 4: 3 hours streaming, 3.5 GB data
Day 5: 1.5 hours streaming, 2 GB data

---

Step 2: Notice a pattern. It seems for every hour of streaming, your data usage increases by about 1 GB, plus a base usage of around 0.5 GB (even if you don't stream much).

---

Step 3: Formulate a simple model. Let 'D' be data usage and 'H' be streaming hours. A simple model could be: D = 1 * H + 0.5

---

Step 4: Test the model with your data:
Day 1: D = 1*1 + 0.5 = 1.5 GB (Matches!)
Day 2: D = 1*2 + 0.5 = 2.5 GB (Matches!)
Day 3: D = 1*0.5 + 0.5 = 1 GB (Matches!)

---

Step 5: Use the model to predict. If you plan to stream for 4 hours tomorrow, your predicted data usage would be: D = 1*4 + 0.5 = 4.5 GB.

Answer: Based on this simple statistical model, you can predict you'll use 4.5 GB of data if you stream for 4 hours.

Why It Matters

Statistical modelling is the backbone of many exciting fields like Artificial Intelligence and Data Science, helping computers learn from data. It's used by scientists to predict weather, economists to forecast market trends, and engineers to design safer systems. Learning this can open doors to careers in AI development, financial analysis, or even sports analytics.

Common Mistakes

MISTAKE: Thinking a model is always 100% accurate and will predict perfectly. | CORRECTION: Statistical models are approximations based on past data; they help make educated guesses but are not perfect crystal balls. There's always some uncertainty.

MISTAKE: Using too little data to build a model. | CORRECTION: Models built with very little data might not be reliable. More data generally leads to better, more robust models that capture patterns accurately.

MISTAKE: Assuming the relationships in the model will never change. | CORRECTION: The world changes, and so do patterns. A model built on old data might become less accurate over time. Models need to be updated and re-evaluated regularly.

Practice Questions

Try It Yourself

QUESTION: A small chai shop tracks its sales. On a rainy day, they sell 50 cups. On a sunny day, they sell 100 cups. What kind of day (rainy or sunny) would you predict if they sold 75 cups? | ANSWER: This is between rainy and sunny, suggesting a partially cloudy or mixed weather day, or perhaps a day with light rain.

QUESTION: A student observes that for every 10 minutes they study, their test score improves by 2 marks. If they studied for 60 minutes, and their base score (with no studying) is 40 marks, what would their predicted test score be? Show your working. | ANSWER: Improvement = (60/10) * 2 = 6 * 2 = 12 marks. Predicted score = Base score + Improvement = 40 + 12 = 52 marks.

QUESTION: You are tracking the price of onions (in Rs/kg) in your local market. For the past 4 weeks, the prices were: Week 1: Rs 20, Week 2: Rs 22, Week 3: Rs 24, Week 4: Rs 26. If this linear trend continues, what would be the predicted price of onions in Week 6? | ANSWER: The price increases by Rs 2 each week. So, Week 5: Rs 28, Week 6: Rs 30. The predicted price in Week 6 would be Rs 30/kg.

MCQ

Quick Quiz

Which of the following is the main purpose of statistical modelling?

To perfectly predict every future event without error.

To understand patterns in data and make informed predictions.

To replace human decision-making entirely.

To only describe past events without looking forward.

The Correct Answer Is:

Statistical modelling helps us find hidden patterns in data to make educated guesses or predictions, not perfect ones. It aids decision-making and goes beyond just describing the past.

Real World Connection

In the Real World

Many apps you use daily, like Swiggy or Zomato, use statistical models to predict how long your food delivery will take, based on traffic, restaurant prep time, and rider availability. Even banks use these models to assess credit risk for loans or to detect fraud in UPI transactions.

Key Vocabulary

Key Terms

DATA: Facts or figures collected for analysis | PATTERN: A regular, understandable way in which something happens or is done | PREDICTION: A forecast or guess about a future event | RELATIONSHIP: How two or more things are connected or affect each other | TREND: A general direction in which something is developing or changing

What's Next

What to Learn Next

Now that you understand what statistical modelling is, the next step is to learn about different types of data and how to visualize them using graphs. This will help you identify patterns more easily before building a model.