Explain the decision tree algorithm, explain with 2 examples.
Write about Ridge regression and lasso regression.
Examine the importance of machine learning frame work for any application btl-3
Examine the importance of machine learning algorithm btl4
Discuss the difference between training set and testing set
Draw the basic learning system model in spam Emil detection
Explain below dataset validation with one example k folds validation, leave out cross validation
1.
A complex model predicts house prices perfectly
on training data but fails on new listings. Is this high bias or high variance?
Explain the trade-off and name one specific technique to fix this issue.
2.
A rare disease classifier has 99% accuracy, but
doctors find it misses most sick patients. Why is accuracy misleading? Which
metric like Precision, Recall, F1should you prioritize, and what does that mean
for false positives?
3.
You need to predict if a customer will purchase
Yes/No based on their browsing time. Why is Linear Regression a poor choice
here? What does Logistic Regression output that makes it suitable, and how is
the final class determined?
4.
A bank uses a Decision Tree for loan approvals.
At a node, it can split applicants by ‘Age>30’ or ‘Income>50k.’ Define
Information Gain. Which split would the algorithm choose, and how does this
relate to reducing impurity?
5.
Your K-NN music recommender works poorly after
adding a new ‘song length’ feature measured in seconds, while ‘genre’ is a
category code. Why did performance drop? What single preprocessing step is
critical before using K-NN, and why?
1.
A
healthcare company wants to predict whether a patient has a specific disease
based on symptoms and medical history. They have a labeled dataset from past
patients. Which type of machine learning would you choose? Justify your choice
and describe how the model would learn from the data.
2.
An
e-commerce platform wants to recommend products to users without any prior
labels on user preferences. They only have data on user browsing history and
purchase logs. Which ML approach would you use? Explain how the algorithm would
discover patterns.
3.
A
robotics engineer is training a robot to navigate a maze. The robot learns by
trial and error, receiving rewards for moving closer to the exit and penalties
for hitting walls. Which ML paradigm fits this problem? Name the key components
such as agent, environment in this context.
4.
You
are given a dataset of house features like size, location, rooms and their
prices. Your task is to predict the price of a new house. Is this a
classification or regression problem? Why? What would be the label in this
dataset?
5.
A
supermarket wants to segment its customers into groups for targeted marketing.
They have data on customer purchase frequency, basket size, and product
categories, but no predefined customer labels. Which unsupervised learning
technique would you apply? How would you evaluate the results?
6.
A
streaming service wants to classify movies into genres automatically based on
their subtitles and metadata. They have a training set with movies already
labeled by genre. Outline the steps of how a supervised model would learn to
classify a new movie.
7.
A
self-driving car must learn to avoid collisions in real-time by observing
traffic, pedestrian movement, and road signs. It receives positive feedback for
safe driving and negative for near misses. How does reinforcement learning
apply here? What would be the state and action in this case?
8.
An
agricultural researcher wants to identify different crop types from satellite
images. They have images labeled as “wheat,” “corn,” or “soybean.” Which type
of supervised learning task is this? What kind of algorithm might be suitable
and why?
9.
A
financial institution wants to detect fraudulent transactions. They have
historical data with labels “fraud” or “not fraud.” Explain how supervised
learning would work here, including what the model learns during training and
how it makes predictions.
10.
A
music app wants to create personalized playlists by grouping songs with similar
audio features like tempo, key, energy. There are no pre-defined categories.
Which unsupervised learning method would you use? Describe how the algorithm
would create song clusters.
Contd.............
కామెంట్లు లేవు:
కామెంట్ను పోస్ట్ చేయండి