patterns in data (“data mining”) that may not be evident initially.
•
Principal Components
•
Clustering Methods
–
K Means
–
Hierarchical
–
Bayesian

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
– a
nonparametric algorithm that can be used for
classification and regression.

K-Nearest Neighbor
Parametric vs. Non-parametric:
Parametric:
Non-parametric:

K-Nearest Neighbor
Parametric vs. Non-parametric:
Parametric: Assumes a specific distribution for
variable(s).
Non-parametric: Does not make any assumptions
with regard to the distribution of variable(s).

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
– a
nonparametric algorithm that can be used for
classification and regression.
•
Objective: To classify a binary random variable in a test
set as 0 or 1 based comparisons to the training data.

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
– a
nonparametric algorithm that can be used for
classification and regression.
•
Objective: To classify a binary random variable in a test
set as 0 or 1 based comparisons to the training data.
•
How: By calculating the distance between k nearest
training observations for a given test observation, and
assigning that observation as either 0 or 1 based on a
vote of the corresponding Y (0 or 1)
values of the k
nearest observations.

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
•
Logic of the KNN Algorithm:
–
KNN is based on feature similarity – i.e., how closely
features resemble our training set determines how we
classify a given data point.

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
•
Logic of the KNN Algorithm:
–
KNN is based on feature similarity – i.e., how closely
features resemble our training set determines how we
classify a given data point.
–
An object is classified by a majority vote of its neighbors,
with the object being assigned to the class most
common among its
k
nearest neighbors
.

K-Nearest Neighbor
K-Nearest Neighbor: “The KNN Classifier”
•
Logic of the KNN Algorithm:
–
KNN is based on feature similarity – i.e., how closely

#### You've reached the end of your free preview.

Want to read all 30 pages?

- Fall '19