#### Unit 1: Capstone Project

1. What is a Capstone Project?

A) A small homework task
B) A final project using all your learning
C) A quiz on AI
D) A coding competition
Answer: B

2. Which is NOT a goal of the Capstone Project?

A) Solve realworld problems
B) Explain solutions in simple words
C) Avoid using algorithms
D) Apply AI learning
Answer: C

3. What is the first step in an AI project?

A) Build a model
B) Understand the problem
C) Collect data
D) Test the model
Answer: B

4. If there’s no pattern in the data, what should you do?

A) Use AI anyway
B) Don’t use AI
C) Build a random model
D) Ignore the data
Answer: B

5. Which question does Classification answer?

A) How much?
B) Which category?
C) Which group?
D) Is this unusual?
Answer: B

6. What does RMSE stand for?

A) Root Mean Squared Error
B) Random Model Standard Error
C) Real Mean Simple Error
D) Root Model Squared Error
Answer: A

7. Which project idea involves text analysis?

A) Stock Prices Predictor
B) Sentiment Analyzer
C) Movie Ticket Price Predictor
 D) Human Activity Recognition
Answer: B

8. How many steps are there in the AI Project Cycle?

A) 4
B) 5
C) 6
D) 7
Answer: C

9. What is the last step in the AI Project Cycle?

A) Data gathering
B) Deployment
C) Model construction
D) Evaluation
Answer: B

10. Which method helps break down complex problems?

A) Design Thinking
B) Random Guessing
C) Trial and Error
D) Ignoring the problem
Answer: A

#### Design Thinking & Problem Decomposition

11. How many stages are in Design Thinking?
A) 3
B) 4
C) 5
D) 6
Answer: C

12. What is the first stage of Design Thinking?

A) Define
B) Empathize
C) Ideate
D) Test
Answer: B

13. What’s the purpose of breaking down a problem?

A) Make it harder
B) Solve it stepbystep
C) Avoid coding
D) Confuse the team
Answer: B

14. In problem decomposition, what should you know first?

 A) The code to use
B) Inputs and outputs
C) The final result
D) The team size
Answer: B

15. What’s an example of a small coding task?

A) Building an app
B) Calculating one book’s volume
C) Predicting weather for a year
D) Designing a website
Answer: B

16. What does the Analytic Approach decide?

A) The team size
B) The method to solve the problem
C) The project deadline
D) The data source
Answer: B

17. Which question fits a Predictive approach?

A) What happened?
B) What will happen next?
C) Why is this happening?
D) How do we solve it?
Answer: B

18. What’s the purpose of a training set?

A) To test the model
B) To build the model
C) To collect data
D) To deploy the model
Answer: B

19. What’s a common split for traintest data?

A) 50%50%
B) 80%20%
C) 90%10%
D) 60%40%
Answer: B

20. What does crossvalidation do?

A) Speeds up model training
B) Tests on different data chunks
C) Reduces data size
D) Simplifies coding
Answer: B
21. When is crossvalidation most useful?
A) Large datasets
B) Small datasets
C) Fast models
D) Simple problems
Answer: B

22. What’s a downside of crossvalidation?

A) Less accurate
B) Takes longer
C) Uses less data
D) Harder to code
Answer: B

23. What library is used to split data in Python?

A) Pandas
B) Sklearn
C) Numpy
D) Matplotlib
Answer: B

24. What does "test_size=0.2" mean?

A) $20\%$ for training
B) $20\%$ for testing
C) $80\%$ for testing
D) $80\%$ for training
Answer: B

25. What’s MAE stand for?

A) Mean Absolute Error
B) Model Accuracy Error
C) Mean Average Error
D) Main Absolute Error
Answer: A

26. What’s a component of time series data?

A) Color
B) Trend
C) Speed
D) Size
Answer: B

27. What does "Seasonality" mean in time series?

A) Random changes
B) Repeating cycles
C) Longterm increase
 D) Average value
Answer: B

28. What dataset is used to show time series decomposition?

A) Housing prices
B) Airline passengers
C) Stock prices
D) Movie tickets
Answer: B

29. What does "Noise" mean in time series?

A) Random variation
B) Seasonal pattern
C) Trend direction
D) Average level
Answer: A

30. What model is assumed for airline passengers data?

A) Additive
B) Multiplicative
C) Linear
D) Random
Answer: B

31. What does RMSE measure?

A) Model speed
B) Prediction error
C) Data size
D) Code length
Answer: B

32. A good RMSE value is:

A) High
B) Low
C) Zero
D) Negative
Answer: B

33. What’s the formula for RMSE?

A) Square root of mean squared error
B) Sum of errors
C) Average of predictions
D) Square of differences
Answer: A

34. What does MSE penalize more?

A) Small errors
 B) Large errors
C) Random errors
D) No errors
Answer: B

35. When should you use MSE?

A) For classification
B) For regression
C) For clustering
D) For anomaly detection
Answer: B

36. What’s the range of MSE?

A) 0 to 1
B) 0 to infinity
C) 1 to 1
D) Negative infinity to infinity
Answer: B

37. What library calculates MSE in Python?

A) Numpy
B) Sklearn
C) Pandas
D) Matplotlib
Answer: B

38. What’s an example of a regression problem?

A) Classifying cats and dogs
B) Predicting house prices
C) Grouping customers
D) Detecting fraud
Answer: B

39. What does a loss function measure?

A) Model speed
B) Prediction accuracy
C) Data size
D) Code complexity
Answer: B

40. What’s gradient descent used for?

A) Maximizing errors
B) Minimizing loss
C) Splitting data
D) Plotting graphs
Answer: B
#### Mixed Questions
41. What’s a key concept in Capstone Projects?
A) Model validation
B) Web design
C) Game coding
D) Data deletion
Answer: A

42. Which stage comes after “Ideate” in Design Thinking?

A) Empathize
B) Prototype
C) Define
D) Test
Answer: B

43. What’s the purpose of a test dataset?

A) Train the model
B) Evaluate the model
C) Collect data
D) Define features
Answer: B

44. What does a Random Forest Regressor do?

A) Classifies data
B) Predicts numbers
C) Clusters groups
D) Detects anomalies
Answer: B

45. What’s an example of anomaly detection?

A) Predicting sales
B) Finding unusual patterns
C) Grouping data
D) Recommending movies
Answer: B

46. What Python library plots data?

A) Sklearn
B) Pandas
C) Matplotlib
D) Numpy
Answer: C

47. What’s a sign of a bad model?

A) Low RMSE
B) High RMSE
C) Zero MSE
 D) Fast training
Answer: B

48. What’s the benefit of cross validation over train test split?
A) Faster
B) More accurate
C) Less data needed
D) Simpler code
Answer: B

49. What’s a feature in machine learning?

A) The output
B) The input data
C) The model type
D) The error rate
Answer: B

50. What’s the goal of storytelling through data?

A) Confuse the audience
B) Explain patterns clearly
C) Hide trends
D) Reduce data size
Answer: B

### 10 Short Questions and Answers

1. What is a Capstone Project?

Answer: It’s a final project where students solve realworld problems using all their
AI knowledge.

2. Name the first step in the AI Project Cycle.

Answer: Understanding the problem.

3. What’s the purpose of Design Thinking?

Answer: To solve complex problems with a solution based approach.

4. What’s a training set used for?

Answer: To build and train the AI model.

5. What does RMSE tell us?

Answer: How far the model’s predictions are from the real values.

6. What’s one component of time series data?

Answer: Trend (or Seasonality, Noise, Level).
7. Why do we split data into train and test sets?
Answer: To train the model on one part and test it on another.

8. What’s the difference between MSE and RMSE?

Answer: MSE is the average of squared errors; RMSE is its square root.

9. What’s cross validation?

Answer: A method to test a model on multiple data chunks for better accuracy.

10. Give an example of a regression problem.

Answer: Predicting house prices.

### 10 Long Questions and Answers

1. Explain the six steps of the AI Project Cycle.

Answer: The AI Project Cycle has six steps:
1. Problem Definition: Understand the problem and check if there’s a pattern.
2. Data Gathering: Collect the data needed to solve it.
3. Feature Definition: Pick the important parts of the data to use.
4. AI Model Construction: Build the AI model using the data.
5. Evaluation & Refinements: Test the model and fix any issues.
6. Deployment: Use the model in the real world. These steps ensure a structured
approach to creating AI solutions.

2. What is Design Thinking, and how does it help in problem solving?

Answer: Design Thinking is a 5stage method to solve problems creatively:
Empathize (understand users), Define (state the problem), Ideate (brainstorm ideas),
Prototype (make solutions), and Test (try them out). It helps by focusing on the user’s
needs, breaking down complex issues, and testing ideas step-by-step to find the best
solution.

3. How do you break down a complex problem into smaller parts? Give an example.
Answer: To break down a problem:
1. Understand it and restate it simply.
2. Split it into big pieces.
3. Break those into smaller, manageable tasks.
4. Code and test each task.
Example: To create an app, decide its purpose (e.g., a weather app), design its
look, pick features (like temperature display), code each part, and test it separately
before combining.

4. What’s the difference between a training set and a test set? Why are both
important?
Answer: A training set is data used to build and teach the model, while a test set is
separate data used to check how well the model works. Both are important because
the training set helps the model learn patterns, and the test set shows if it can predict
accurately on new, unseen data, ensuring it’s not just memorizing.

5. Describe the train test split method and how it’s done in Python.
Answer: Train test split divides data into two parts: one to train the model (e.g.,
80%) and one to test it (e.g., 20%). In Python, using
`sklearn.model_selection.train_test_split`, you load data (e.g., with pandas), split it
into features (X) and labels (y), then use `train_test_split(X, y, test_size=0.2)` to get
`X_train`, `X_test`, `y_train`, and `y_test`.

6. What is cross validation, and why is it better than a simple traintest split?
Answer: Cross validation tests a model by splitting data into multiple folds (e.g., 5),
using each fold as a test set while training on the rest, then averaging the results. It’s
better than traintest split because it uses all data for testing, reducing randomness
and giving a more reliable measure of model quality, especially for small datasets.

7. Explain RMSE with an example of how it’s calculated.

Answer: RMSE (Root Mean Squared Error) measures prediction error by finding the
square root of the average squared differences between predicted and actual values.
Example: If actual values are [1, 2, 3] and predicted are [1.5, 2.2, 2.8], differences are
[0.5, 0.2, 0.2], squared are [0.25, 0.04, 0.04], mean is 0.11, and RMSE is √0.11 ≈ 0.33.

8. How does MSE work, and when should it be used?

Answer: MSE (Mean Squared Error) is the average of squared differences between
predicted and actual values, penalizing bigger errors more. It’s used in regression
problems (e.g., predicting prices) when you assume data is normally distributed and
want to heavily penalize outliers. Formula: MSE = Σ(predicted actual)² / n.

9. What are the components of time series data? Give an example from the airline
passengers dataset.
Answer: Time series data has four components:
1. Level: Average value.
2. Trend: Longterm increase/decrease.
3. Seasonality: Repeating cycles.
4. Noise: Random changes.
Example: In the airline passengers dataset (19491960), there’s a rising trend
(more passengers over time), seasonality (peaks each year), and noise (random
fluctuations).

10. Discuss a Capstone Project idea like the Sentiment Analyzer and how you’d
approach it.
Answer: A Sentiment Analyzer predicts if text (e.g., reviews) is positive, negative,
or neutral. Approach:
1. Problem: Define it as classifying text emotions.
2. Data: Gather reviews or tweets.
3. Features: Use words or phrases as inputs.
4. Model: Build a classification model (e.g., with sklearn).
 5. Test: Split data, train, and check accuracy with metrics like MSE.
6. Deploy: Use it to analyze new text. This integrates AI skills to solve a practical
problem.