.Rhistory

# Find the path of working directory
getwd()
# Set the path of working directory to Cardiohealthpython
setwd("C:\\Users\\User\\OneDrive\\Documents\\Cardiohealthpython")
dir()
# Load necessary libraries
library(dplyr)
library(ggplot2)
library(stats)
# Unmask packages
stats::filter
stats::lag
base::intersect
base::setdiff
base::setequal
base::union
conflicts()
cvd_dataset <- read.csv("updated_cardio_health_multi1.csv")
# Display the first few rows of the data
head(cvd_dataset)
# Create a data frame
df <- data.frame(cvd_dataset)
table(df$cardio_condition, df$smoker)
chisq.test(table(df$cardio_condition, df$smoker))
b. Drinking
table(df$cardio_condition, df$alcohol)
chisq.test(table(df$cardio_condition, df$alcohol))
table(df$cardio_condition, df$active)
chisq.test(table(df$cardio_condition, df$active))
table(df$cardio_condition, df$glucose_levels)
chisq.test(table(df$cardio_condition, df$glucose_levels))
table(df$cardio_condition, df$cholesterol_levels)
chisq.test(table(df$cardio_condition, df$cholesterol_levels))
table(df$smoker, df$alcohol)
chisq.test(table(df$smoker, df$alcohol))
table(df$smoker, df$active)
chisq.test(table(df$smoker, df$active))
table(df$alcohol, df$active)
chisq.test(table(df$alcohol, df$active))
t.test(body_mass_index ~ cardio_condition, data = df)
# Visualizations
# Create boxplots for each variable
ggplot(df, aes(x = cardio_condition, y = body_mass_index, fill = cardio_condition)) +
geom_boxplot() +
labs(title = "BMI by CVD Status", x = "Presence of CVD", y = "Body Mass Index")
# Create bar plots for categorical variables (cardio_condition vs lifestyle habits)
ggplot(df, aes(x = cardio_condition, fill = smoker)) +
geom_bar(position = "fill") +
labs(title = "Smoking by CVD Status", x = "Presence of CVD", y = "Proportion", fill = "Smoker")
# Create bar plots for categorical variables (cardio_condition vs alcohol)
ggplot(df, aes(x = cardio_condition, fill = alcohol)) +
geom_bar(position = "fill") +
labs(title = "Smoking by CVD Status", x = "Presence of CVD", y = "Proportion", fill = "Alcohol")
# Create bar plots for categorical variables (cardio_condition vs alcohol)
ggplot(df, aes(x = cardio_condition, fill = alcohol)) +
geom_bar(position = "fill") +
labs(title = "Alcohol Consumption by CVD Status", x = "Presence of CVD", y = "Proportion", fill = "Alcohol")
# Create bar plots for categorical variables (cardio_condition vs active)
ggplot(df, aes(x = cardio_condition, fill = active)) +
geom_bar(position = "fill") +
labs(title = "Physical Activity by CVD Status", x = "Presence of CVD", y = "Proportion", fill = "Physical Activity")