AN 300 EXAM 2 WITH A GUARANTEED PASS!!
Cluster Analysis Answer - -It is a descriptive analytics technique
-It is to answer what has happened questions.
-It is used to discover natural groupings of objects
Problem Characteristics of Cluster Analysis Answer - -The data that describes
the object must be given.
-Its objective is to maximize similarities of objects within groups.
-It is about how to organize objects into groups.
-No prior knowledge of how the objects are related to each other
Euclidean distance Answer - -For Numerical data
-the distance between two objects is the length of a straight line between them
-It requires replacing the raw value of data with its z-score(=(raw value-
mean)/standard deviation))
-The lower the measure, the better
-Ex: Height in inches, weight in pounds
Matching coefficient Answer - -For Categorical data
-It is a ratio of number of columns with matching categorical values to the total
number of categorical columns.
-Ex: gender and income levels of customers
, Business Applications of Cluster Analysis Answer - Marketing- find target
market groups
Finance- create groups for personalized financial advice
Operations- identifying outliers for quality control
Similarity Measures Answer - gauge whether a group of objects are similar or
dissimilar to one another
Hierarchical Clustering Answer - -use for small sets of data (less than 500 rows)
-supports numeric and categorical data
-sensitive to outliers
-experiment with different methods to calculate the distance between clusters
K-Means Clustering Answer - -useful for large data sets (more than 500 rows)
-supports only numeric data
-less sensitive to outliers
-experiment with different number of clusters
Cubic Clustering criterion (CCC) Answer - -a metric related to R^2, the
proportion of variance in the data accounted for by the clusters
-It is used to tell the goodness of fit of clustering
-A value greater than two indicated good clusters
-A value between 0 and 2 indicates possible clustering
-large negative numbers of CCC (ex. -30) may be due to outliers
Hierarchical Clustering Process Answer - 1) starts with placing each object into
its own cluster
Cluster Analysis Answer - -It is a descriptive analytics technique
-It is to answer what has happened questions.
-It is used to discover natural groupings of objects
Problem Characteristics of Cluster Analysis Answer - -The data that describes
the object must be given.
-Its objective is to maximize similarities of objects within groups.
-It is about how to organize objects into groups.
-No prior knowledge of how the objects are related to each other
Euclidean distance Answer - -For Numerical data
-the distance between two objects is the length of a straight line between them
-It requires replacing the raw value of data with its z-score(=(raw value-
mean)/standard deviation))
-The lower the measure, the better
-Ex: Height in inches, weight in pounds
Matching coefficient Answer - -For Categorical data
-It is a ratio of number of columns with matching categorical values to the total
number of categorical columns.
-Ex: gender and income levels of customers
, Business Applications of Cluster Analysis Answer - Marketing- find target
market groups
Finance- create groups for personalized financial advice
Operations- identifying outliers for quality control
Similarity Measures Answer - gauge whether a group of objects are similar or
dissimilar to one another
Hierarchical Clustering Answer - -use for small sets of data (less than 500 rows)
-supports numeric and categorical data
-sensitive to outliers
-experiment with different methods to calculate the distance between clusters
K-Means Clustering Answer - -useful for large data sets (more than 500 rows)
-supports only numeric data
-less sensitive to outliers
-experiment with different number of clusters
Cubic Clustering criterion (CCC) Answer - -a metric related to R^2, the
proportion of variance in the data accounted for by the clusters
-It is used to tell the goodness of fit of clustering
-A value greater than two indicated good clusters
-A value between 0 and 2 indicates possible clustering
-large negative numbers of CCC (ex. -30) may be due to outliers
Hierarchical Clustering Process Answer - 1) starts with placing each object into
its own cluster
