Intro To Classification

Question 1

What is classification?

Answer 1

The prediction of objects by class

Question 2

What is binary classification?

Answer 2

The model is making a prediction between two classes, example dog or cat and yes or no

Question 3

What is multiclass classification?

Answer 3

A model prediction between multiple classes, example: meat, dairy, fruit or vegetables

Question 4

What is balanced or unbalanced classification?

Answer 4

Evaluating how balanced the classes are such as both sides have equal amount if dogs and cats. The balance of dataset will help us interpret of classification

Question 5

What is decision trees for classification?

Answer 5

Works similar to Regression except final prediction is not an average but the class itself, also called classification trees

Question 6

What is false positives? (Type 1 errors)

Answer 6

Also known as false positives is a common error type, for example phrase like want to find employees but avoid bad hires(false positives) even turning away qualified people(false negatives)

Question 7

What is False negatives? (Type 2 errors)

Answer 7

Is another common error Type where we predict that something is negative when it is in fact, positive. Example a false negative for a covid test but in reality the person does have covid

Question 8

What is a confusion matrix?

Answer 8

A table that is often used to describe the performance of a classification model (or “classifier”) on a set of test data for which the true values are known.

Question 9

What is accuracy?

Answer 9

Accuracy is the most intuitive metric, in other worlds accuracy is correct predictions our model made out of the total number of precitions

Question 10

What is Recall?

Answer 10

When we want to reduce the number of false negatives, we want to improve recall

Question 11

What is Precision?

Answer 11

When we want to reduce the numbers of the false positives, we want to improve precision

Question 12

What is class probabilities?

Answer 12

Calculates the probabilities of each class and returns the most likely

Question 13

What is Decision Thresholds?

Answer 13

When working with a binary classification problem, the model prediction will, by default, be the class whoe probability is greater than 0.5 or 50%

Question 14

What is K Nearest Neighbors (KNN)?

Answer 14

K-nearest neighbors is a model that uses the “k” most similar observations in order to make a prediction.

Question 15

What are the Advantages of KNN?

Answer 15

Easier to understand and explain than other machine learning algorithms. Can be used for classification or Regression or can be used for multi-class classification problems

Question 16

What are the Disadvantages of KNN?

Answer 16

It must be store all of the data it was fit on. It’s prediction phase can be slow when the data it was “fit” on is large. Typically worse performance than other supervised learning methods

Question 17

What is Logistic Regression?

Answer 17

Logistic Regression uses something called a sigmoid function which bounds the output between 0 and 1

Question 18

What is Cross-Validation?

Answer 18

Cross-validation creates and trains multiple identical models using different subsets of the data

Question 19

What is Tuning Hyperparameters?

Answer 19

Choosing a set of optimal hyperparameters for learning algorithm. A hyperparameter is a model argument whose value is set before the learning process begins

Question 20

What is definition of ROC curve?

Answer 20

A plot of true positive rate (y axis) vs the false positive rate (x axis) for every possible classification threshold

Question 21

What is the point of regularization?

Answer 21

To reduce variance

Question 22

What is another name for ‘L1’ Regularization?

Answer 22

Lasso

Question 23

What do we want to do with the cost function?

Answer 23

Minimize the cost function