What is Underfitting? | Simple Explanation for Class 10

S3-SA5-0444

What is Underfitting on a Graph?

Grade Level:

Class 10

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

Definition

What is it?

Underfitting on a graph happens when your model is too simple and cannot capture the main trend or pattern in the data. It's like trying to draw a complex rangoli design with only a few straight lines – you miss all the beautiful curves and details. The model performs poorly on both the data it learned from and new, unseen data.

Simple Example

Quick Example

Imagine you are trying to predict the price of a plate of samosas based on the day of the week. If you only use a straight line (a very simple model) to connect the prices, but the prices actually go up on weekends and down on weekdays, your line will not show this pattern. It will underfit, meaning your predictions will be far off for most days.

Worked Example

Step-by-Step

Let's say we have data for the number of ice creams sold (y-axis) versus the temperature (x-axis):

Temperature (Celsius): 20, 22, 25, 28, 30
Ice Creams Sold: 10, 15, 25, 35, 40

---Step 1: Plot these points on a graph. You'll see the points generally go upwards in a slightly curved way.

---Step 2: Now, let's try to fit a very simple model – a horizontal straight line, like 'Ice Creams Sold = 25'. This line suggests the number of ice creams sold is always 25, no matter the temperature.

---Step 3: Look at the graph. Does this horizontal line accurately represent the increasing trend? No, it clearly misses the pattern that more ice creams are sold as temperature rises.

---Step 4: This horizontal line is an example of underfitting. It's too simple and doesn't learn the relationship between temperature and ice cream sales at all.

---Answer: The horizontal line 'Ice Creams Sold = 25' underfits the data because it fails to capture the clear upward trend.

Why It Matters

Understanding underfitting is crucial in fields like AI/ML and Data Science because it helps engineers build accurate predictive models. If a model underfits, it gives wrong predictions, which can be critical in areas like medical diagnosis or financial forecasting. Data Scientists and AI Engineers use this knowledge to choose better models for tasks like predicting crop yields or traffic patterns.

Common Mistakes

MISTAKE: Thinking underfitting means the model is too complex. | CORRECTION: Underfitting actually means the model is too SIMPLE and cannot capture the underlying patterns in the data.

MISTAKE: Confusing underfitting with overfitting. | CORRECTION: Underfitting is when the model is too simple and performs poorly on both training and new data. Overfitting is when the model is too complex, learns the training data too well (including noise), and performs poorly on new data.

MISTAKE: Believing underfitting only happens with straight lines. | CORRECTION: While straight lines are common examples, any model that is too basic for the complexity of the data can underfit, even if it's not a simple straight line.

Practice Questions

Try It Yourself

QUESTION: A student tries to predict their monthly mobile data usage based on the number of hours they watch videos. They use a single average value for all months. Is this an example of underfitting? | ANSWER: Yes, this is likely an example of underfitting. Using a single average value is too simple and won't capture how video watching hours directly affect data usage, especially if video watching varies a lot.

QUESTION: You are plotting the growth of a plant over several weeks. The plant initially grows slowly, then rapidly, then slows down again (an 'S' shape). If you try to fit a straight line to this data, what problem would you likely face? | ANSWER: You would likely face underfitting. A straight line is too simple to capture the 'S' shaped growth pattern, missing the initial slow growth, the rapid middle growth, and the final slowing down.

QUESTION: An economist is trying to model the price of petrol in India over the last year. They use a model that only considers the average global crude oil price and ignores factors like government taxes, exchange rates, and local demand. Explain why this model might underfit the actual petrol price data. | ANSWER: This model might underfit because it is too simplistic. Petrol prices are influenced by many complex factors besides just global crude oil prices, such as government taxes, the value of the Indian Rupee against the dollar (exchange rates), and local demand fluctuations. By ignoring these crucial factors, the model cannot capture the true variations and patterns in petrol prices, leading to poor predictions.

MCQ

Quick Quiz

Which of the following best describes underfitting on a graph?

The model is too complex and learns too much from the training data.

The model is too simple and fails to capture the main patterns in the data.

The model performs perfectly on new, unseen data.

The model has too many features and becomes slow.

The Correct Answer Is:

Underfitting occurs when the model is too simple and cannot learn the underlying relationship or pattern in the data, leading to poor performance on both known and new data. Options A describes overfitting, while C and D are incorrect descriptions.

Real World Connection

In the Real World

In cricket analytics, if a model tries to predict a batsman's score based only on the temperature during the match, it would likely underfit. A batsman's score depends on many factors like pitch condition, bowler's skill, past performance, and pressure. A simple temperature-only model would fail to capture these complexities, giving poor predictions, which is why actual cricket analytics use far more advanced models.

Key Vocabulary

Key Terms

MODEL: A simplified representation of a real-world process or system. | PATTERN: A regular and intelligible form or sequence discernible in the way data points are arranged. | PREDICTION: A forecast or guess about a future event or outcome. | DATA: Facts and statistics collected together for reference or analysis.

What's Next

What to Learn Next

Now that you understand underfitting, you should learn about 'Overfitting on a Graph'. It's the opposite problem, where a model becomes too complicated. Knowing both will help you understand how to build models that are 'just right'!