Frosi Exam: Chapter 3

이 룰렛의 항목 (44개)

• What is classification?
• What is the output of: logistic regression and classification? How are they encoded?
• What is logistic regression? Explain and write the model
• Majority of data follows non-linear behavior, how can you solve the non-linearity problem?
• Polynomial model of the second degree can be obtained from what model?
• How is classification different from regression?
• Which is the method of classification we learned in class?
• What are the cost and confusion matrix? Explain them.
• If we find out that a binary classification and algorithm is giving 99.9% accuracy should we trust this value? Why not?
• What is precision and recall?
• What is the K-nearest neighbor. Explain by plotting the graph. What is the problem with KNN model? Are there any parameters like weights in Linear Regression associated with KNN? Tell me one possible way to calculate K? Which is the type of method used in practical lessons to evaluate hyperparameters?
• What are the counterparts of logistic regression?
• Dendrogram. What is it?
• What is an outlier? FORMAL DEFINITION.
• Can we use logistic regression for categorical data? What would you use?
• What is a cost matrix? What is a confusion matrix? How many elements does a cost matrix have and why?
• Perfect idea classifier. Do you remind how this translates to precision curve? Rock curve?
• What happens when the precision curve gets close to 1?
• Name the phenomenal that occurs with noise?
• What is a cost matrix? How many elements does a cost matrix have? Why four? Cuz you use it with confusion matrix.
• Do you remember the idea classifier has a precision recall curve? 100% precision independent from recall.
• What happens to the precision and recall matrix when the threshold used to _ classification gets to 1? And what is the opposite? What is the trick to make the model linear?
• What is the shape of the logistics function? What are the boundaries? Imagine now that you have a very simple logistic function in which the model is a linear model, why is it w0 and w1. How does w0 and w1 influence the shape of the logistic function?
• What are the 2 types of classifications?
• How do we predict the labels for linear classifiers?
• What is a decision tree? How is it different graphical from logistic regression?
• What is Nearest Neighbors classifier?
• To decide the label for an unseen data via Nearest Neighbor classification you need what?
• How do you classify using Nearest Neighbor?
• Draw K-Nearest Neighbours Classifier (1-k, 2-k and 3-k).
• Why is k-Nearest considered a lazy learner?
• What is the function used to compute the similarity between 2 examples?
• How would you determine the class from nearest neighbor?
• How would you normalize classifiers? For nominal features? Missing values?
• How would you evaluate classification algorithm?
• What is the equation for accuracy?
• Why would a model predict everything to be 99.9% Why is this misleading?
• How can we fix a misleading accuracy prediction of 99%?
• How do you compute cost matrix? Confusion matrix?
• What is precision and recall and F-1 measure? Write the equations.
• Suppose we use a near one threshold to classify positive examples? What happens to precision and recall?
• Suppose we use a near zero threshold to classify positive examples? What happens to precision and recall?
• Draw a ROC curve. What would happen if we were to change the classification threshold or sample distribution or cost matrix? What do the data points mean?
• Draw a precision recall curve. What does the best classifier look like?