anna-data-science-blog - Tumblr blog

[anna-data-science-blog]

Week 4: Graphing the Data

*********************

CODE

*********************

""" Spyder Editor

Creating Graphs For Your Data """

import pandas import numpy import seaborn import matplotlib.pyplot as plt

#read in data set in form of CSV full_data = pandas.read_csv('nesarc_pds.csv', low_memory=False)

data = full_data[['S2BQ1B1', 'S13Q6A1', 'AGE']].copy()

#frequency distributions will be ordered according to numeric values data['S2BQ1B1'] = pandas.to_numeric(data['S2BQ1B1'], errors='coerce') data['S13Q6A1'] = pandas.to_numeric(data['S13Q6A1'], errors='coerce')   data['AGE'] = pandas.to_numeric(data['AGE'], errors='coerce')

# Data trimmed to people over the age of 40 sub1=data[(data['AGE']>=40)]

sub2=sub1.copy()

#convert "Unknown" to NAN sub2["S2BQ1B1"] = sub2["S2BQ1B1"].replace(9, numpy.nan) sub2["S13Q6A1"] = sub2["S13Q6A1"].replace(9, numpy.nan)

#relabel responses to reflect "yes" or "no" sub2["ALCOHOLISM_IN_PAST_12_MOS"] = sub2["S2BQ1B1"].astype('category') sub2["ALCOHOLISM_IN_PAST_12_MOS"] = sub2["ALCOHOLISM_IN_PAST_12_MOS"].cat.rename_categories(["yes", "no"])

# Convert "no"(2) to 0 sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"] = sub2["S13Q6A1"].astype('category') sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"] = sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"].cat.rename_categories([1, 0]) sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"] = pandas.to_numeric(sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"], errors="coerce")

#cut data into 40-49, 50-59, 60-69, 70+ sub2['AGE_GROUP'] = pandas.cut(data.AGE, [39, 50, 60, 70, 120]) sub2['AGE_GROUP'] = sub2['AGE_GROUP'].astype('category') sub2['AGE_GROUP'] = sub2['AGE_GROUP'].cat.rename_categories(["40-49 y/o", "50-59 y/o", "60-69 y/o", "70+ y/o"])

#univariate bar chart for alcoholism categories seaborn.countplot(x="ALCOHOLISM_IN_PAST_12_MOS", data=sub2) plt.xlabel("Alcoholism in the past 12 months") plt.title("Alcholism in the past 12 months among individuals >= 40y/o in the NESARC Study")

#univariate bar chart for arteriosclerosis categories seaborn.countplot(x="ARTERIOSCLEROSIS_IN_PAST_12_MOS", data=sub2) plt.xlabel("Arteriosclerosis in the past 12 months") plt.title("Arteriosclerosis in the past 12 months among individuals >= 40y/o in the NESARC Study")

#univariate bar chart for age groups of respondents seaborn.countplot(x="AGE_GROUP", data=sub2) plt.xlabel("Age groups") plt.title("Age Groups of Respondents >= 40y/o in the NESARC Study")

#a bivariate graph for alcoholism in the past 12 months(x-axis) and atherosclerosis in the past 12 months (y-axis) seaborn.factorplot(x="ALCOHOLISM_IN_PAST_12_MOS", y="ARTERIOSCLEROSIS_IN_PAST_12_MOS", data=sub2, kind="bar", ci=None) plt.xlabel("Alcoholism in the past 12 months") plt.ylabel("Proportion of Arteriosclerosis in the past 12 months")

*********************

GRAPHS & SUMMARIES

*********************

The univariate graph of alcoholism in the past 12 months:

This graph is unimodal, with most observations falling under the category of “no” for alcoholism in the past 12 months for respondents 40 years of age or older. 

The univariate graph of arteriosclerosis in the past 12 months:

This graph is unimodal, with most observations falling under the category of “no” (0) for arteriosclerosis in the past 12 months for respondents 40 years of age or older. 

The univariate graph of age groups of respondents:

This graph is unimodal, with the highest peak at the age group of 40-49 years old. It can be considered right-skewed, as there are higher frequencies in the lower age ranges.

The bivariate graph of alcoholism in the past 12 months vs. the proportion of arteriosclerosis in the past 12 months:

According to this graph, of the respondents aged 40y/o+ who experienced alcoholism in the past 12 months, approximately 2.3% experienced arteriosclerosis in the past 12 months. Of those who did not experience alcoholism, approximately 2.6% experienced arteriosclerosis in the past 12 months. In short, there was less than a 0.5% difference in rates of arteriosclerosis between those who did and those who did not experience alcoholism in the previous year. 

[anna-data-science-blog]

Week 3: Data Management

"""

CODE

"""

#!/usr/bin/env python3

# -*- coding: utf-8 -*-

"""

Spyder Editor

Data Management

"""

import pandas

import numpy

#read in data set in form of CSV

full_data = pandas.read_csv('nesarc_pds.csv', low_memory=False)

data = full_data[['S2BQ1B1', 'S13Q6A1', 'AGE']].copy()

"""

#number of individuals

print("number of participants")

print(len(data))

"""

#frequency distributions will be ordered according to numeric values

data['S2BQ1B1'] = pandas.to_numeric(data['S2BQ1B1'], errors='coerce')

data['S13Q6A1'] = pandas.to_numeric(data['S13Q6A1'], errors='coerce')  

data['AGE'] = pandas.to_numeric(data['AGE'], errors='coerce')

# Data trimmed to people over the age of 40

sub1=data[(data['AGE']>=40)]

sub2=sub1.copy()

#convert "Unknown" to NAN

sub2["S2BQ1B1"] = sub2["S2BQ1B1"].replace(9, numpy.nan)

sub2["S13Q6A1"] = sub2["S13Q6A1"].replace(9, numpy.nan)

#relabel responses to reflect "yes"(1) or "no" (2)

sub2["ALCOHOLISM_IN_PAST_12_MOS"] = sub2["S2BQ1B1"].astype('category')

sub2["ALCOHOLISM_IN_PAST_12_MOS"] = sub2["ALCOHOLISM_IN_PAST_12_MOS"].cat.rename_categories(["yes", "no"])

sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"] = sub2["S13Q6A1"].astype('category')

sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"] = sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"].cat.rename_categories(["yes", "no"])

#cut data into 40-49, 50-59, 60-69, 70+

sub2['AGE_GROUP'] = pandas.cut(data.AGE, [39, 50, 60, 70, 120])

sub2['AGE_GROUP'] = sub2['AGE_GROUP'].astype('category')

sub2['AGE_GROUP'] = sub2['AGE_GROUP'].cat.rename_categories(["40-49 y/o", "50-59 y/o", "60-69 y/o", "70+ y/o"])

#count & percentages: individuals over or equal to 40 years old

print("counts for AGE")

c3 = sub2["AGE_GROUP"].value_counts(sort=False).sort_index()

print(c3)

print("percentages for AGE")

p3 = sub2["AGE_GROUP"].value_counts(sort=False, normalize=True).sort_index()

print(p3)

#count & percentages: alcohol abuse in the past 12 months

print("counts for alcohol abuse in the past 12 months, where age >=40")

c4 = sub2["ALCOHOLISM_IN_PAST_12_MOS"].value_counts(sort=False, dropna=False)

print(c4)

print("percentages for alcohol abuse in the past 12 months, where age >=40")

p4 = sub2["ALCOHOLISM_IN_PAST_12_MOS"].value_counts(sort=False, normalize=True, dropna=False)

print(p4)

#count & percentages: arteriosclerosis in the past 12 months

print("counts for arteriosclerosis in the past 12 months, where age >=40")

c5 = sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"].value_counts(sort=False, dropna=False)

print(c5)

print("percentages for arteriosclerosis in the past 12 months, where age >=40")

p5 = sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"].value_counts(sort=False, normalize=True, dropna=False)

print(p5)

print("alcoholism in the past 12 months, by age group")

print(pandas.crosstab(sub2["AGE_GROUP"], sub2["ALCOHOLISM_IN_PAST_12_MOS"]))

print("arteriosclerosis in the past 12 months, by age group")

print(pandas.crosstab(sub2["AGE_GROUP"], sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"]))

print("arteriosclerosis and alcoholism in the past 12 months")

print(pandas.crosstab(sub2["ALCOHOLISM_IN_PAST_12_MOS"], sub2["ARTERIOSCLEROSIS_IN_PAST_12_MOS"]))

"""

OUTPUT

"""

"""

SUMMARY

"""

For this week’s assignment, I analyzed three variables: the ages of respondents, whether or not they experienced alcoholism in the past 12 months, and whether or not they experienced arteriosclerosis in the past 12 months. My data management included grouping the different respondents into age groups to make analysis more manageable, replacing “Unknown” or “NA” responses with NaN, and renaming categories to be more legible (“yes” or “no” rather than “1” or “2”).

I performed cross-tabulations for all three of my variables.

For age groups, respondents between 40-49 years old — the youngest in the current group studied— most commonly responded that they experienced alcoholism in the past months.  On the other hand, respondents 70+ years old most frequently responded that they experienced arteriosclerosis in the past 12 months.

[anna-data-science-blog]

Running My First Program

******************************************

MY PROGRAM

******************************************

# -*- coding: utf-8 -*- """ Spyder Editor

This is my first program, performing exploratory analysis on the NESARC data for the question: Is there an association between alcohol abuse in the past 12 months and ateriosclerosis in the past 12 months, for individuals greater than or equal to 40 years old? """

import pandas import numpy

#read in data set in form of CSV data = pandas.read_csv('nesarc_pds.csv', low_memory=False)

#frequency distributions will be ordered according to numeric values data['S2BQ1B1'] = pandas.to_numeric(data['S2BQ1B1'], errors='coerce') data['S13Q6A1'] = pandas.to_numeric(data['S13Q6A1'], errors='coerce')   data['AGE'] = pandas.to_numeric(data['AGE'], errors='coerce')

# Data trimmed to people over the age of 40 sub1=data[(data['AGE']>=40)]

sub2=sub1.copy()

#count & percentages: individuals over or equal to 40 years old print("counts for AGE") c3 = sub2["AGE"].value_counts(sort=False).sort_index() print(c3)

print()

print("percentages for AGE") p3 = sub2["AGE"].value_counts(sort=False, normalize=True).sort_index() print(p3)

#count & percentages: alcohol abuse in the past 12 months print("counts for alcohol abuse in the past 12 months, where age >=40") c4 = sub2["S2BQ1B1"].value_counts(sort=False) print(c4)

print("percentages for alcohol abuse in the past 12 months, where age >=40") p4 = sub2["S2BQ1B1"].value_counts(sort=False, normalize=True) print(p4)

#count & percentages: arteriosclerosis in the past 12 months print("counts for arteriosclerosis in the past 12 months, where age >=40") c5 = sub2['S13Q6A1'].value_counts(sort=False) print(c5)

print("percentages for arteriosclerosis in the past 12 months, where age >=40") p5 = sub2['S13Q6A1'].value_counts(sort=False, normalize=True) print(p5)

******************************************

FREQUENCY TABLES

******************************************

counts for AGE 40    992 41    881 42    912 43    832 44    823 45    878 46    799 47    810 48    789 49    742 50    729 51    732 52    693 53    711 54    742 55    594 56    589 57    541 58    621 59    502 60    503 61    500 62    454 63    458 64    453 65    448 66    442 67    423 68    405 69    464 70    426 71    423 72    397 73    432 74    441 75    384 76    367 77    377 78    317 79    297 80    324 81    262 82    233 83    237 84    175 85    175 86    139 87    149 88     94 89     98 90     76 91     53 92     45 93     26 94     24 95     16 96     13 97      9 98     14 Name: AGE, dtype: int64

percentages for AGE 40    0.038925 41    0.034569 42    0.035786 43    0.032647 44    0.032294 45    0.034452 46    0.031352 47    0.031783 48    0.030959 49    0.029115 50    0.028605 51    0.028723 52    0.027192 53    0.027899 54    0.029115 55    0.023308 56    0.023112 57    0.021228 58    0.024367 59    0.019698 60    0.019737 61    0.019619 62    0.017814 63    0.017971 64    0.017775 65    0.017579 66    0.017344 67    0.016598 68    0.015892 69    0.018207 70    0.016716 71    0.016598 72    0.015578 73    0.016951 74    0.017304 75    0.015068 76    0.014401 77    0.014793 78    0.012439 79    0.011654 80    0.012713 81    0.010281 82    0.009143 83    0.009300 84    0.006867 85    0.006867 86    0.005454 87    0.005847 88    0.003688 89    0.003845 90    0.002982 91    0.002080 92    0.001766 93    0.001020 94    0.000942 95    0.000628 96    0.000510 97    0.000353 98    0.000549 Name: AGE, dtype: float64

counts for alcohol abuse in the past 12 months, where age >=40 2.0    13790 1.0      404 9.0      214 Name: S2BQ1B1, dtype: int64

percentages for alcohol abuse in the past 12 months, where age >=40 2.0    0.957107 1.0    0.028040 9.0    0.014853 Name: S2BQ1B1, dtype: float64

counts for arteriosclerosis in the past 12 months, where age >=40 1      881 2    23757 9      847 Name: S13Q6A1, dtype: int64

percentages for arteriosclerosis in the past 12 months, where age >=40 1    0.034569 2    0.932195 9    0.033235 Name: S13Q6A1, dtype: float64

******************************************

SUMMARY

******************************************

A random sample of 43,093 were asked to report their ages. Of all these respondents, 25,485 (59.1%) were greater than or equal to 40 years old.

The participants were asked if alcohol abuse/dependence had happened for them in the last 12 months. Of all of the respondents 40 years old or greater, 2.8 % chose category 1 (yes), 1.5 % chose category 9, indicating the question was not applicable of the individual was a lifetime abstainer, and the remaining majority were under category 2.0 (no).

For the next question, the same participants were asked if they experienced arteriosclerosis or hardening of the arteries in the past 12 months. Of all of the participants aged 40+, 3.5 % fell under category 1 (yes), 3.3% where under category 9 [Unknown], and the remaining 93.2 % chose category 2 (no).

[anna-data-science-blog]

Starting the Research Project

As an engineer with a medical background, I was particularly interested in studying the NESARC data, and in particular the possible associations between alcohol abuse and coronary conditions, as both are serious chronic illnesses in the country where I live. I am particularly interested in the exploring the association of whether alcohol abuse/dependence happened in the last 12 months and associating that whether the subject experienced hardening of arteries or arteriosclerosis in the last 12 months, as I have known people with comorbidity of the two conditions. To summarize my current research question: Is alcohol abuse in the last 12 months associated with arteriosclerosis in the last 12 months?

As such, I have added to my codebook variables reflecting alcohol abuse and variables reflecting cardiovascular health within the same time frame. I performed a review of literature, and several articles remarked that alcohol abuse is a known risk factor for cardiovascular disease. The authors of “Markers of early vascular ageing”* list alcohol as a classic risk factor alongside “age, gender, diabetes mellitus, dyslipidemia.” According to the article “White coat hypertension may be an initial sign of accelerated atherosclerotic process,”* alcohol intake may even have a causative relationship to accelerated atherosclerosis—a condition known as metabolic syndrome.

My hypothesis is that there will be a positive correlation between alcohol abuse and the hardening of arteries in the last 12 months.

Works Cited

* Kotsis, V., et al. "Markers of early vascular ageing." Current pharmaceutical design (2017).

* Helvaci, Mehmet Rami, et al. "White coat hypertension may be an initial sign of an accelerated atherosclerotic process." Middle East Journal of Internal Medicine 10.2 (2017).