WorksheetsIntermediate Level (10 Questions)
Total questions: 12
Worksheet time: 7mins
Scenario: You are developing a system for a logistics company to predict the estimated time of arrival (ETA) for delivery trucks. Using Mitchell’s framework, how would you define the Task (T)?
The historical GPS logs of all delivery trucks.
The average error in minutes between predicted and actual arrival.
Predicting the arrival time for a specific delivery route.
Improving the route efficiency over a period of six months.
A bank wants to automate its credit card approval process. They provide you with thousands of past applications labeled as "Approved" or "Rejected." Which type of learning is most appropriate here?
Unsupervised Learning.
Supervised Learning: Classification.
Reinforcement Learning.
Supervised Learning: Regression.
In a real estate application, you are trying to predict the exact price of a house based on its square footage and location. This is an example of:
Regression.
Clustering.
Classification.
Association.
You are analyzing a dataset of customer shopping baskets. You find that "80% of customers who buy bread also buy jam." This is a classic example of:
Supervised classification.
Reward-based reinforcement.
Unsupervised association.
Continuous regression.
An AI agent is learning to play a video game by receiving "points" for staying alive and "penalties" for crashing. This interaction with an environment to maximize a reward is called:
Supervised Learning.
Unsupervised Learning.
Data Mining.
Reinforcement Learning.
Which of the following is the most accurate definition of "Experience (E)" for a handwriting recognition system?
The percentage of digits correctly recognized by the software.
The actual task of identifying a "7" from a "1".
A database of handwritten images paired with their correct labels.
The computational power of the processor running the algorithm.
What is a primary niche for Machine Learning, as identified by Mitchell?
Tasks that are too complex to be programmed by hand, like autonomous driving.
Situations where data is limited and computational power is low.
Designing hardware for better data storage.
Calculating simple arithmetic operations in a spreadsheet.
An industrial sensor provides noisy data. Your model perfectly fits every single data point in the training set, including the anomalies. On the test set, however, the accuracy drops significantly. This is a sign of:
Low bias and low variance.
High bias (underfitting).
High variance (overfitting).
An unbiased learner.
According to Mitchell, what would happen to a learner that has no "Inductive Bias"?
It would become more efficient at searching the hypothesis space.
It could only classify instances it has previously seen and could not generalize to new data.
It would automatically find the most specific hypothesis (S).
It would require significantly fewer training examples to reach high accuracy.
A "Version Space" is formally defined as:
The set of all hypotheses that are consistent with the training data.
The space of all possible instances in the dataset.
The transition from the most specific to the most general hypothesis.
The set of all attributes that have a "?" constraint.
If a business owner asks why your model's predictions are often slightly off, you explain that you chose a simpler model (like a straight line) instead of a complex one to ensure it generalizes better. This error due to simplified assumptions is called:
Variance.
Irreducible Error.
Bias.
Overfitting.
Of the following examples, which would you address using an unsupervised learning algorithm? check all that apply
Given a database of customer data, automatically discover market segments and group customers into different market segments
Given email labeled as spam/not spam, learn a spam filter
Given a dataset of patients diagnosed as either having diabetes or not, learn to classify new patients as having diabetes or not
Given a set of news articles found on the web, group them into set of articles about the same story
