COMPARING ASSOCIATION RULES AND DECISION TREES FOR
Download
Report
Transcript COMPARING ASSOCIATION RULES AND DECISION TREES FOR
COMPARING ASSOCIATION
RULES AND DECISION TREES
FOR DISEASE PREDICTION
Carlos Ordonez
MOTIVATION
Three main issues about mining association rules
in medical datasets:
1.
2.
3.
A significant fraction of association rules is
irrelevant
Most relevant rules with high quality metrics
appear only at low support
# of discovered rules becomes extremely large at
low support
Search constraints:
Find only medically significant association rules
Make search more efficient
MOTIVATION
Decision tree a well-known machine learning
algorithm
Association rules vs. Decision tree
Accuracy
Interpretability
Applicability
ASSOCIATION RULES
Support
Confidence
Lift
confidence( x y )
lift ( x y )
sup port ( x y )
sup port ( x)
confidence( x y )
sup port ( y )
Lift quantifies the predictive power of x y
Rules such that lift(xy) > 1 are interesting!
CONSTRAINED ASSOCIATION RULES
Transforming Medical Data Set
Data must be transformed to binary dimensions
Numeric attributes intervals, each interval is mapped to
an item.
Categorical attributes each categorical value is an item
If an attribute has negation add that as an item
Each item is corresponds to the presence or absence of one
categorical value or one numeric interval
CONSTRAINED ASSOCIATION RULES
1.
Search Constraints
Max itemset size (k)
2.
Group
3.
Reduces the combinatorial explosion of large
itemsets and helps finding simple rules
gi >0 Aj belongs to a group
gi =0 Aj is not group-constrained at all
This avoids finding trivial or redundant rules
Antecedent/Consequent
ci = 1 Ai is an antecedent
ci = 2 Ai is a consequent
Binned at 40(adult) and
60(old)
Percentage of vessel narrowing
LAD, LCX and RCA are binned
at 70% and 50%
LM is binned at 30% and 50%
9 heart regions ( 2 ranges
with 0.2 as cutoff)
Patients
655
attributes
25
Binned at 200 and 250
PARAMETERS
k=4
Min support = 1% ≈ 7
Min confidence = 70%
Min lift = 1.2
To get rules where there is stronger implication
dependence between X and Y
Rules with conf ≥ 90 and lift ≥ 2, with 2 or more
items in the consequent were considered
medically significant.
HEALTHY ARTERIES
9,595 associations
771 rules
DISEASED ARTERIES
Several unneeded items
were filtered out ( with
values in lower (healthy)
ranges)
10,218 associations
552 rules
PREDICTIVE RULES FROM DECISION
TREES
CN4.5 using gain ratio
CART similar results
Threshold for the height of the tree to produce
simple rules
Percentage of patients (ls)
Fraction of patients where the antecedent appears
Confidence factor (cf)
Focus on predicting LDA disease
PREDICTIVE RULES FROM DECISION
TREES
1.
All measurements without binning as
independent variables, numerical variables are
automatically split
Without any threshold on height:
181 node
90% accuracy
height = 14
most rules more than 5 attributes
except 5 rules, other involve less than 2% of the patients
More than 80% of rules refer to less than 1% of patients
Many rules involve attributes with missing information
Many rules had the same variable being split several times
Few rules with cf = 1 but splits included borderline cases
and involves few patients
PREDICTIVE RULES FROM DECISION
TREES
With threshold = 10 on height
83 nodes
77% accuracy
Most rules have repeated attributes
More than 5 attributes
Perfusion cutoffs higher than 0.5
Low cf and involved less than 1% of the population
With threshold = 3 on height
65% accuracy
Simpler rules
RELATED WORK
PREDICTIVE RULES FROM DECISION
TREES
2.
Items (binary variables) as independent
variables like association rules are used
With threshold = 3 on height
Most of the rules were much closer to the prediction
requirements
10 nodes
DISCUSSION
Decision trees
are not as powerful as association rules in this case
Do not work well with combinations of several target
variables
Fail to identify many medically relevant
combinations of independent numeric variable ranges
and categorical values
Tend to find complex and long rules, if the height is
unlimited
Find few predictive rules with reasonably sized (>1%)
sets of patients in such cases
Rules some times repeat the same attribute
DISCUSSION - ALTERNATIVES
build many decision trees with different
independent attributes
Create a family of small trees, each tree has a
weight
It’s error-prone, difficult to interpret, slow for higher
# of attributes
Each tree becomes similar to a small set of
association rules
Constraints for association rules can be adopted
to decision trees (future work)
DISCUSSION – DECISION TREE
ADVANTAGES
DT partitions the data set, ARs on the same
target attributes may refer to overlap
DT represents a predictive model of data set, ARs
are disconnected among themselves
DT is guaranteed to have at least 50% prediction
accuracy and generally above 80% for binary
target variables, ARs require trial and error to
find the best threshold