Problems #1 (20 points)
Using an “Addiction” dataset, a researcher has prepared the following table of patient counts:
Ethnicity
Age Category
Alcohol
Cocaine
Heroin
Row Total
Black
Old
30
48
17
95
Young
25
72
13
110
Hispanic
Old
7
0
5
12
Young
8
7
19
34
White
Old
60
2
17
79
Young
26
10
34
70
Column Total
156
139
105
400
Use the table above and Excel to classify patient addiction type (Alcohol, Cocaine, Heroin) using Ethnicity and Age Category:
• Construct a C4.5 decision tree (two levels deep only).
Problem #2 (15 points)
Use R/Python and the NYNJ_zipcode_population.csv file to cluster the zip codes of New Jersey and New York into 3 clusters. Use “Low_income” percent and “Total_Pop” as clustering attributes. (Algorithm=hierarchical, linkage=simple).
• What are the members of each cluster?
• Is there an unusual zip code? If so, what is the zip code?
Problem #3 (10 points)
Use R/Python and the NYNJ_zipcode_population.csv file to cluster the zip codes of New Jersey and New York into 5 clusters. Use “Low_income” percent and “Total_Pop” as clustering attributes (Algorithm=kmeans).
• What are the members of each cluster?
Problem #4 (20 points)
Use an Artificial Neural Network, with 10 nodes in the hidden layer, learning rate of .001, and the standard scaler (StandardScaler) to develop a classification model for the heart_attack .CSV dataset using 30% of the data as training data. (See the data dictionary below).
• What is the model accuracy ?
Problem #5 (20 points)
Use Random Forest to develop a classification model for the heart_attack .CSV dataset using 30% of the data as training data. (See the data dictionary below).
• What is the most important feature? Why?
• Show your model metrics.
Problem #6 (15 points)
Using data in the table below, construct a Neural Network with one Output Layer (z) and one Hidden Layer (A and B).
• Calculate the predicted outcome if the inputs to the input nodes are (x=1, Node 1=.4, Node 2=.7 Node 3= .7 and Node 4=.2).
• Adjust the weight if the actual output is 0.8500
From
To
Weight
X
A
0.5
Node 1
A
0.6
Node 2
A
0.8
Node 3
A
0.6
Node 4
A
0.2
x
B
0.7
Node 1
B
0.9
Node 2
B
0.8
Node 3
B
0.4
Node 4
B
0.2
xx
z
0.5
A
z
0.9
B
z
0.9
Data Dictionary
Attribute
Description
Target/Attribute
Heart_Attack
Type of Heart Attack (Mild, Light, Massive)
Target
RestHR
Resting Heart Rate
Attribute
MaxHR
Maximum Heart Rate
Attribute
RecHR
Recovery Heart Rate
Attribute
BP
Blood Pressure
Attribute
Data dependencies: NYNJ_zipcode_population.csv, Heart_attack.CSV
