Week 3

Question 1

Nearest Neighbor Algorithm

Answer 1

• Nearest Neighbor classifiers are defined by their characteristics of classifying unlabeled cases based on the similarity of labeled cases
• for example, if user A and B exhibit the same purchasing behavior the items purchased by A are also recommended to user B.
• Nearest neighbor algorithms are used for
  
  • recommendations
  • identifying patterns in data
• suited when the data type is homogenous

Question 2

K-NN

Answer 2

• k-NN stands for K-Nearest Neighbor Algorithm
• the simple machine learning algorithm

Question 3

Strengths of KNN

Answer 3

• simple and effecctive
• makes no assumptions about the underlying data distribution
• fast training phase

Question 4

Weakness of KNN

Answer 4

• does not produce a model, limiting the ability to understand how the feature is related to the class
• requires selection of an appropriate k
• slow classification phase
• nominal features and missing data required additional processing

Question 5

Letter K in K-NN Algorithm

Answer 5

• the letter is K is a variable term is implying the numbers of neighbors that could be used.
• for each unlabeled record in the test dataset, k-NN identifies k records in the training data that are the “nearest” in similarity

Question 6

Euclidean Distance

Answer 6

• Euclidean distance is used to measure the similarity between two instances
• where p and q are the examples to be compared
• each having n features
• the term p1 refers to the value of the first feature of example p, while q1 refers to the value of the first feature of example q

Question 7

Choosing a value for K in K-NN

Answer 7

• k value determines how well the model will generalize to future data
• the balance between over lifting and underlighting the training data is a problem known as bias-variance tradeoff
• choosing a large k reduces the impact or variance caused by noisy data, but can bias the learner so that it runs the risk of ignoring small but important patterns
• choosing smaller values for K can be noisy and will have a higher influence on the result.

Question 8

commonly used techniques

Answer 8

k=sqrt(N) where N stands for the number of samples in your training dataset