Big data : CM3
Cluster (groupe) analysis groups data : information found in
data (stocks, people, region, products, etc )
Hierarchical clustering : set of
nested clusters organized as a
Clustering methods hierarchical tree
Partitional clustering : division of
dta objects into non-overlapping
subsets such
Why partitional ? Why Hierarchical ?
natural complement to the PCA More suitable for
algorithlm categorical data
Easier to understand and interpret
Powerful methd for business
applications
No knowledge of a
Scalable : suitable and cheap response variable is
Fast required
Unlabelled data
K-means Simple
Flexible Exploits distance
clustering
Interpretable measures
Partitional
Observations withing clusters are closer to
Clustering : exploratory, each other
descriptive, data analysis Observations across clusters are further
away
Big data : CM3 1
Cluster (groupe) analysis groups data : information found in
data (stocks, people, region, products, etc )
Hierarchical clustering : set of
nested clusters organized as a
Clustering methods hierarchical tree
Partitional clustering : division of
dta objects into non-overlapping
subsets such
Why partitional ? Why Hierarchical ?
natural complement to the PCA More suitable for
algorithlm categorical data
Easier to understand and interpret
Powerful methd for business
applications
No knowledge of a
Scalable : suitable and cheap response variable is
Fast required
Unlabelled data
K-means Simple
Flexible Exploits distance
clustering
Interpretable measures
Partitional
Observations withing clusters are closer to
Clustering : exploratory, each other
descriptive, data analysis Observations across clusters are further
away
Big data : CM3 1
