Week 4 Assignmemt: Graphing Decisions

NESARC Dataset: Questions Selected

As part of this assignment I will be considering association between the frequency of drinking alcohol during the time of abuse (:S2AQ21A: Explanatory Variable) and whether or not the person suffered from one of the below cardiovascualar medical condition in last 12 months:

  • Hardening of Arteries (S13Q6A1: Response Variable)
  • High Blood Pressure (S13Q6A2: Response Variable)
  • Heat Attack (S13Q6A7: Response Variable)
All four variables are categorical.

Univariate Bar Graps

Since all variables are categorical, I have used Bar Graphs for Univariate plots. Below are the bar graphs for the four variables:







Descriptions for the selected variables

I have used the Pandas describe() method to describe the variables. Below is the output of the Python program:

Total rows in NESARC dataset: 43093
Total columns in NESARC dataset: 3010
Description for how frequently people drank alcohol during the time of alcohol abuse: S2AQ21A
count    34331.000000
unique      10.000000
top          1.000000
freq      5215.000000
Name: S2AQ21A, dtype: float64
Description for if people suffered from hardening of arteries in last 12 months: S13Q6A1
count    41828.000000
unique       2.000000
top          0.000000
freq     40917.000000
Name: S13Q6A1, dtype: float64
Description for if people suffered from high blood pressure in last 12 months: S13Q6A2
count    41964.000000
unique       2.000000
top          0.000000
freq     32828.000000
Name: S13Q6A2, dtype: float64
Description for if people suffered from heart attack in last 12 months: S13Q6A7
count    42027.000000
unique       2.000000
top          0.000000
freq     41557.000000
Name: S13Q6A7, dtype: float64


Bivariate Graphs

I have used the seaborn.catplot() to plot the association between the explanatory variable and the three response variables. Below are the bivariate bar graphs for the same:






Python Code

Below is the Python / Pandas code used to generate above plots:

import pandas
import numpy
import seaborn
import matplotlib.pyplot as plt
 
#Set PANDAS to show all columns in DataFrame
pandas.set_option('display.max_columns', None)
#Set PANDAS to show all rows in DataFrame
pandas.set_option('display.max_rows', None)
 
# bug fix for display formats to avoid run time errors
pandas.set_option('display.float_format', lambda x:'%f'%x)
 
# Explanatory variable
colFreqAlcoholDuringAbuse = 'S2AQ21A'
 
# response variables
colHardeningOfArteries = 'S13Q6A1'
colHighBloodPressure = 'S13Q6A2'
colHeartAttack = 'S13Q6A7'
 
data = pandas.read_csv('nesarc_pds.csv', low_memory=False)
 
print (f"Total rows in NESARC dataset: {len(data)}") #number of observations (rows)
print (f"Total columns in NESARC dataset: {len(data.columns)}") # number of variables (columns)
 
#setting variables you will be working with to numeric (updated)
data[colFreqAlcoholDuringAbuse] = pandas.to_numeric(data[colFreqAlcoholDuringAbuse], errors='coerce')
data[colHardeningOfArteries] = pandas.to_numeric(data[colHardeningOfArteries], errors='coerce')
data[colHighBloodPressure] = pandas.to_numeric(data[colHighBloodPressure], errors='coerce')
data[colHeartAttack] = pandas.to_numeric(data[colHeartAttack], errors='coerce')
 
# Dara Management Step 1: Recode UNKNOWN (99 or 9) to NaB for response variables
data[colFreqAlcoholDuringAbuse] = data[colFreqAlcoholDuringAbuse].replace(99, numpy.nan)
data[colHardeningOfArteries] = data[colHardeningOfArteries].replace(9, numpy.nan)
data[colHighBloodPressure] = data[colHighBloodPressure].replace(9, numpy.nan)
data[colHeartAttack] = data[colHeartAttack].replace(9, numpy.nan)
 
# Data Management Step 2: Recode NO (2) for response variable to 0
data[colHardeningOfArteries] = data[colHardeningOfArteries].replace(2, 0)
data[colHighBloodPressure] = data[colHighBloodPressure].replace(2, 0)
data[colHeartAttack] = data[colHeartAttack].replace(2, 0)
 
# Data Management Step3: Reversing alcohol drinking frequency categorierec
# currently 1 is most frequent (every day in a week) and 10 is least frequent (once or twice in a year). Therefore reversing the order
recode = {1:10, 2:9, 3:8, 4:7, 5:6, 6:5, 7:4, 8:3, 9:2, 10:1}
data[colFreqAlcoholDuringAbuse] = data[colFreqAlcoholDuringAbuse].map(recode)
 
# Make all the variables as categorical
data[colFreqAlcoholDuringAbuse] = data[colFreqAlcoholDuringAbuse].astype('category')
data[colHardeningOfArteries] = data[colHardeningOfArteries].astype('category')
data[colHighBloodPressure] = data[colHighBloodPressure].astype('category')
data[colHeartAttack] = data[colHeartAttack].astype('category')
 
# UNnivariate bar graphs for categorical variables
freqAlcoholPlot = seaborn.countplot(x=colFreqAlcoholDuringAbuse, data=data)
plt.xlabel('Alcohol Drinking Frequwncy')
plt.title('Frequency of drinking alcohol during alcohol abuse')
fig = freqAlcoholPlot.get_figure()
fig.savefig('AlcoholFrequencyBarChart.png')
 
hardeningOfArteriesPlot = seaborn.countplot(x=colHardeningOfArteries, data=data)
plt.xlabel('People suffered Hardening of Arteries')
plt.title('People who suffered Hardening of Arteries in last 12 months')
fig = hardeningOfArteriesPlot.get_figure()
fig.savefig('ArteriesHardeningBarChart.png')
 
highBPPlot = seaborn.countplot(x=colHighBloodPressure, data=data)
plt.xlabel('People suffered High Blood Pressure')
plt.title('People who suffered High Blood Pressure in last 12 months')
fig = highBPPlot.get_figure()
fig.savefig('HighBloodPressureBarChart.png')
 
heartAttackPlot = seaborn.countplot(x=colHeartAttack, data=data)
plt.xlabel('People suffered Heart Attack')
plt.title('People who suffered Heart Attack in last 12 months')
fig = heartAttackPlot.get_figure()
fig.savefig('HeartAttackBarChart.png')
 
# describing the variables
print(f"Description for how frequently people drank alcohol during the time of alcohol abuse: {colFreqAlcoholDuringAbuse}")
desc1 = data[colFreqAlcoholDuringAbuse].describe()
print(desc1)
 
print(f"Description for if people suffered from hardening of arteries in last 12 months: {colHardeningOfArteries}")
desc2 = data[colHardeningOfArteries].describe()
print(desc2)
 
print(f"Description for if people suffered from high blood pressure in last 12 months: {colHighBloodPressure}")
desc3 = data[colHighBloodPressure].describe()
print(desc3)
 
print(f"Description for if people suffered from heart attack in last 12 months: {colHeartAttack}")
desc4 = data[colHeartAttack].describe()
print(desc4)
 
# Bivariate Plots Categorical t0 Categorical Bar Charts
data[colHardeningOfArteries] = pandas.to_numeric(data[colHardeningOfArteries], errors='coerce')
causal_hardeningOfArteries_plot = seaborn.catplot(x=colFreqAlcoholDuringAbuse, y=colHardeningOfArteries, data=data, kind="bar", ci=None)
plt.xlabel('Alcohol Frequency')
plt.ylabel('Hardening of Arteries Proportion ')
# fig = causal_heartattack_plot.get_figure()
plt.savefig('Causal_ArteriesHardening.png')
 
# Bivariate Plots Categorical to Categorical Bar Charts
data[colHighBloodPressure] = pandas.to_numeric(data[colHighBloodPressure], errors='coerce')
causal_highBP_plot = seaborn.catplot(x=colFreqAlcoholDuringAbuse, y=colHighBloodPressure, data=data, kind="bar", ci=None)
plt.xlabel('Alcohol Frequency')
plt.ylabel('High Blood Pressure Proportion ')
# fig = causal_heartattack_plot.get_figure()
plt.savefig('Causal_HighBloodPressure.png')
 
# bivariate bar graph C->Q
data[colHeartAttack] = pandas.to_numeric(data[colHeartAttack], errors='coerce')
causal_heartattack_plot = seaborn.catplot(x=colFreqAlcoholDuringAbuse, y=colHeartAttack, data=data, kind="bar", ci=None)
plt.xlabel('Alcohol Frequency')
plt.ylabel('Heart Attack Proportion ')
# fig = causal_heartattack_plot.get_figure()
plt.savefig('Causal_HeartAttack.png')

Summary

  • The above bivariate plots have frequency of alcohol drinking during alcohol abuse on x-axis and whether or not the person suffered from Hardenng of Arteries, High Blood Pressure or Heart Attack in last 12 months on the y-axis.
  • On x-axis as we move right, the frequency of drinking alcohol increases from 1 being once or twice a year to 10 being every day of the week.
  • On the y-axis for the three bivariate plots, we have the proportion of the survey candidates who suffered from the three cardiovascular medical condition in last 12 months.
  • As is apparent from the bivariate plots, there doesn't seem to be a direct relationship between frequency of drinking alcohol and a person suffering from hardening of arteries, high blood pressure or heart attack.
  • The results are similar to literatue research performed in Assignment 1.

Comments

Post a Comment

Popular posts from this blog

Data Management and Visualization - Research Topic

Image
Research Dataset Selection - An Association of Alcohol Abuse on Cardio Vascular Health Although not a medical student, human health and what impacts its negatively has always raised my curiosity. Thus, after going through multiple codebook and datasets, I have decided to use in the NESARC study conducted on around 43093 U.S citizens. I would like to study the association between alcohol abuse and occurrence of various Cardio Vascular health conditions. Research Question For my research topic I will be studying the association between alcohol abuse, quantity of abuse, duration of abuse, kind of intoxicant consumed on the occurrence of the following five cardio vascular medical conditions in the last 12 months: Hyper Tension (High Blood Pressure) Arteriosclerosis (Hardening of Artery) Myocardial Infarction (Heart Attack) Tachy Cardia (Rapid Heart Beat) Angina Pectoris (Chest Pain) My Code Book After thorough study of the NESARC codebook, I have selected the following columns from the...
Read more

Week 3 Assignment: Data Management

Variables Selected As part of this assignment, I have selected the following three variables from the NESARC dataset to perform data manipulation operations: Had hardening of arteries in last 12 months (S13Q6A1) Had high blood pressure in last 12 months (S13Q6A2) Had heart attack in last 12 months (S13Q6A7) Python Program to perform Data Management Below is the Python program that I used to perform data management operations on the selected variables: import pandas import numpy # setting options so that pandas data do not get truncated data = pandas.read_csv('nesarc_pds.csv', low_memory=False) print (f"Total rows in NESARC dataset: {len(data)}") #number of observations (rows) print (f"Total columns in NESARC dataset: {len(data.columns)}") # number of variables (columns) # column names colHard...
Read more