Untitled

Research Question

I am interested in analysing impact of variables like urabnisation, female employment, carbon emission and oil consumption per capita on breast cancer in females. All the explanatory variables used have high correlation with stress and environmental pollution. Is there a relationship between increasing stress and air pollution on breast cancer in females?

My Code

# Importing necessary packages import pandas as pd import numpy as np import seaborn as sns import matplotlib.pyplot as plt # importing data df = pd.read_csv("gapminder.csv", index_col = 0, header = 0) # converting required column to numeric for i in df.columns: df[i] = df[i].replace(" ", np.NaN) df[i] = pd.to_numeric(df[i]) #making bins for breast cancer cases df['breastcancerper100bins'] = pd.cut(x = df['breastcancerper100th'], bins = [3, 23, 43, 63, 83, 103]) # making labels for ranges of data df['urbanfactor'] = " " df['urbanfactor'] = np.where(df['urbanrate'] < 60, 'R', df['urbanfactor']) # R means underdeveloped or developing df['urbanfactor'] = np.where((df['urbanrate'] >= 60), 'D', df['urbanfactor']) # D means developed df['urbanfactor'] = np.where(df['urbanrate'] >=80, 'U', df['urbanfactor']) # U means urbanized df['urbanfactor'] = df['urbanfactor'].replace(" ", np.NaN) # # for femaleemployee rate df['femaleemployratebins'] = pd.cut(x = df['femaleemployrate'], bins = [11, 26, 41, 56, 71, 86]) # # for suicideper100th df['suicideper100bins'] = pd.cut(x = df['suicideper100th'], bins = [0, 7, 14, 21, 28, 36]) # #for carbon emmission df['co2emissionsbins'] = pd.qcut(df['co2emissions'], 5) #extracting columns of interest in dataframe sub = df[['breastcancerper100th', 'breastcancerper100bins', 'femaleemployrate','femaleemployratebins','co2emissions','co2emissionsbins', 'suicideper100th', 'suicideper100bins', 'urbanrate', 'urbanfactor']] #univariate distribution plt.figure(1) plt.figure(figsize = (12, 6)) sns.distplot(sub['breastcancerper100th'].dropna()) plt.xlabel('Breast Cancer Rate') plt.ylabel('Frequency') plt.title('Number of new cases of breast cancer in 100,000 female') plt.figure(2) plt.figure(figsize = (12, 6)) sns.distplot(sub['femaleemployrate'].dropna()) plt.xlabel('Female employement Rate') plt.ylabel('Frequency') plt.title('Percentage of female population, age above 15') plt.figure(3) plt.figure(figsize = (12, 6)) sns.distplot(sub['co2emissions'].dropna()) plt.xlabel('Cumulative CO2 emission') plt.ylabel('Frequency') plt.title('Total amount of CO2 emission in metric tons since 1751') plt.figure(4) plt.figure(figsize = (12, 6)) sns.distplot(sub['suicideper100th'].dropna()) plt.xlabel('Suicide, age adjusted, per 100 000') plt.ylabel('Frequency') plt.title('Mortality due to self-inflicted injury, per 100 000') plt.figure(5) plt.figure(figsize = (12, 6)) sns.distplot(sub['urbanrate'].dropna()) plt.xlabel('Urban population') plt.ylabel('Frequency') plt.title('Urban population %age of Total Population') # central tendency measure print('describe Breast Cancer Rate') print(sub['breastcancerper100th'].describe()) print('describe Female employement Rate') print(sub['femaleemployrate'].describe()) print('describe Cumulative CO2 emission') print(sub['co2emissions'].describe()) print('describe Suicide, age adjusted, per 100 000') print(sub['suicideper100th'].describe()) print('describe Urban population') print(sub['urbanrate'].describe()) # factor plots plt.figure(6) plt.figure(figsize = (12, 6)) sns.barplot(sub['urbanfactor'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('urbanfactor') plt.title('Breast Cancer Rate vs urbanfactor') plt.figure(7) plt.figure(figsize = (12, 6)) sns.barplot(sub['femaleemployratebins'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('femaleemployratebins') plt.title('Breast Cancer Rate vs femaleemployratebins') plt.figure(8) plt.figure(figsize = (12, 6)) sns.barplot(sub['co2emissionsbins'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('Cumulative CO2 emission bins') plt.title('Breast Cancer Rate vs Cumulative CO2 emission bins') plt.figure(9) plt.figure(figsize = (12, 6)) sns.barplot(sub['suicideper100bins'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('suicideper100bin') plt.title('Breast Cancer Rate vs suicideper100bin') # regression plot plt.figure(10) plt.figure(figsize = (12, 6)) sns.regplot(sub['urbanrate'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('urbanfactor') plt.title('Breast Cancer Rate vs urbanrate') plt.figure(11) plt.figure(figsize = (12, 6)) sns.regplot(sub['femaleemployrate'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('femaleemployrate') plt.title('Breast Cancer Rate vs femaleemployrate') plt.figure(11) plt.figure(figsize = (12, 6)) sns.regplot(sub['co2emissions'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('co2emissions') plt.title('Breast Cancer Rate vs co2emissions') plt.figure(11) plt.figure(figsize = (12, 6)) sns.regplot(sub['suicideper100th'], sub['breastcancerper100th']) plt.ylabel('Breast Cancer Rate') plt.xlabel('suicideper100th') plt.title('Breast Cancer Rate vs suicideper100thr')

Univariate Analysis

Graphs

Describe Breast Cancer Rate count 173.000000 mean 37.402890 std 22.697901 min 3.900000 25% 20.600000 50% 30.000000 75% 50.300000 max 101.100000 So this is not symmetrical with median near 30, peaks near 20,and 70 and mode near 20

Describe Female employement Rate count 178.000000 mean 47.549438 std 14.625743 min 11.300000 25% 38.725000 50% 47.549999 75% 55.875000 max 83.300003 This is symmetrical with mean, median, mode near 50 Describe Cumulative CO2 emission count 2.000000e+02 mean 5.033261e+09 std 2.573812e+10 min 1.320000e+05 25% 3.484617e+07 50% 1.859018e+08 75% 1.846084e+09 max 3.342210e+11 This is very highly right skewed with mode near 0. It is not symmetrical.

Describe Suicide, age adjusted, per 100 000 count 191.000000 mean 9.640839 std 6.300178 min 0.201449 25% 4.988449 50% 8.262893 75% 12.328551 max 35.752872 This is not symmetrical, right skewed and mean > median.

Describe Urban population count 203.000000 mean 56.769360 std 23.844933 min 10.400000 25% 36.830000 50% 57.940000 75% 74.210000 max 100.000000 Not symmetrical, Bimodal with modes near 30 and 60.

Bivariate analysis

Categorical vs Quantitative

This graph supports our hypothesis as countries with large urban population have more breast cancer cases.

This doesn turned out as expected but this maybe possible due to less data points for countries with higher female employement rate. So, we didnt get any conclusive evidence as above.

This supports out hypothesis as we are getting that increased carbon emission have more breast cancer cases.

This is not turned out as expected maybe be due to beacause this variable doesnt capture the stress as we supposed it to.

Quantitative vs Quantitative Graphs

we can see week positive correlation between these variables supporting our hypothesis

No conclusive evidence

this plot doesnt makes the picture as clear as in the above case

No conclusive evidence

Summary

Hypothesis

Stress increase and exposure to pollution can cause increase breast cancer cases.

Variable of focus in data set:

Response variable = breastcancerper100TH Explanatory variable = femaleemployrate Explanatory variable = co2emissions Explanatory variable = suicideper100TH Explanatory variable = urbanrate In the available dataset unique identifier of a data point is country.

Except the breast cancer variable all other variables are included to capture the features of stress and pollution.