Build and deploy a stroke prediction model using R
Wan Mohamad Hanis bin Wan Hassan
2024-11-05
About Data Analysis Report
This file contains the report of the data analysis done for the project on building and deploying a stroke prediction model in R. It contains analysis such as data exploration, summary statistics and building the prediction models. The final report was completed on Tue Nov 5 03:20:38 2024.
Data Description:
According to the World Health Organization (WHO) stroke is the 2nd leading cause of death globally, responsible for approximately 11% of total deaths.
This data set is used to predict whether a patient is likely to get stroke based on the input parameters like gender, age, various diseases, and smoking status. Each row in the data provides relevant information about the patient.
Task 1: Import data and data preprocessing
Load data and install packages
# Install necessary packages if not already installed
if (!require(caret)) install.packages('caret', dependencies=TRUE)## Loading required package: caret## Loading required package: ggplot2## Loading required package: latticeif (!require(randomForest)) install.packages('randomForest')## Loading required package: randomForest## randomForest 4.7-1.1## Type rfNews() to see new features/changes/bug fixes.##
## Attaching package: 'randomForest'## The following object is masked from 'package:ggplot2':
##
## marginif (!require(e1071)) install.packages('e1071')## Loading required package: e1071if (!require(plumber)) install.packages('plumber')## Loading required package: plumber# Load required libraries
library(caret)
library(randomForest)
library(e1071)
library(plumber)
# Simulate stroke dataset
set.seed(123)
n <- 1000 # Number of observations
# Simulated data
data <- data.frame(
gender = sample(c("Male", "Female"), n, replace = TRUE),
age = sample(25:85, n, replace = TRUE),
hypertension = sample(c(0, 1), n, replace = TRUE),
heart_disease = sample(c(0, 1), n, replace = TRUE),
ever_married = sample(c("Yes", "No"), n, replace = TRUE),
work_type = sample(c("Private", "Self-employed", "Govt_job", "Children", "Never_worked"), n, replace = TRUE),
Residence_type = sample(c("Urban", "Rural"), n, replace = TRUE),
avg_glucose_level = round(runif(n, 50, 300), 1),
bmi = round(runif(n, 10, 50), 1),
smoking_status = sample(c("formerly smoked", "never smoked", "smokes", "Unknown"), n, replace = TRUE),
stroke = sample(c(0, 1), n, replace = TRUE, prob = c(0.9, 0.1)) # 90% non-stroke, 10% stroke
)
# View first few rows of the dataset
head(data)## gender age hypertension heart_disease ever_married work_type
## 1 Male 37 1 0 No Govt_job
## 2 Male 32 1 0 No Never_worked
## 3 Male 26 1 0 No Never_worked
## 4 Female 83 1 1 No Govt_job
## 5 Male 30 0 1 Yes Self-employed
## 6 Female 47 1 0 Yes Govt_job
## Residence_type avg_glucose_level bmi smoking_status stroke
## 1 Rural 204.5 33.6 formerly smoked 0
## 2 Urban 213.4 13.4 formerly smoked 1
## 3 Rural 283.5 40.1 smokes 0
## 4 Rural 202.2 16.2 Unknown 0
## 5 Urban 232.8 12.7 never smoked 0
## 6 Rural 91.4 14.8 never smoked 0# Convert categorical variables to factors
data$gender <- as.factor(data$gender)
data$ever_married <- as.factor(data$ever_married)
data$work_type <- as.factor(data$work_type)
data$Residence_type <- as.factor(data$Residence_type)
data$smoking_status <- as.factor(data$smoking_status)
data$stroke <- as.factor(data$stroke)
# Summary of the dataset
summary(data)## gender age hypertension heart_disease ever_married
## Female:494 Min. :25.00 Min. :0.000 Min. :0.000 No :540
## Male :506 1st Qu.:39.00 1st Qu.:0.000 1st Qu.:0.000 Yes:460
## Median :53.00 Median :0.000 Median :1.000
## Mean :53.46 Mean :0.482 Mean :0.505
## 3rd Qu.:68.00 3rd Qu.:1.000 3rd Qu.:1.000
## Max. :85.00 Max. :1.000 Max. :1.000
## work_type Residence_type avg_glucose_level bmi
## Children :200 Rural:480 Min. : 50.5 Min. :10.00
## Govt_job :177 Urban:520 1st Qu.:112.8 1st Qu.:19.80
## Never_worked :196 Median :171.9 Median :29.60
## Private :204 Mean :172.9 Mean :29.68
## Self-employed:223 3rd Qu.:235.5 3rd Qu.:39.00
## Max. :299.6 Max. :50.00
## smoking_status stroke
## formerly smoked:258 0:899
## never smoked :249 1:101
## smokes :244
## Unknown :249
##
## # Check for missing values
sum(is.na(data))## [1] 0## Describe and explore the dataTask Two: Build prediction models
# Split the data into training and testing sets
set.seed(123)
trainIndex <- createDataPartition(data$stroke, p = 0.7, list = FALSE)
train_data <- data[trainIndex,]
test_data <- data[-trainIndex,]
# Train a Random Forest model
set.seed(123)
rf_model <- randomForest(stroke ~ ., data = train_data, importance = TRUE)
print(rf_model)##
## Call:
## randomForest(formula = stroke ~ ., data = train_data, importance = TRUE)
## Type of random forest: classification
## Number of trees: 500
## No. of variables tried at each split: 3
##
## OOB estimate of error rate: 10.27%
## Confusion matrix:
## 0 1 class.error
## 0 629 1 0.001587302
## 1 71 0 1.000000000# Task Three: Evaluate and select prediction models
# Predict on the test data
rf_predictions <- predict(rf_model, test_data)
# Confusion matrix to evaluate the model
confusionMatrix(rf_predictions, test_data$stroke)## Confusion Matrix and Statistics
##
## Reference
## Prediction 0 1
## 0 269 30
## 1 0 0
##
## Accuracy : 0.8997
## 95% CI : (0.8599, 0.9313)
## No Information Rate : 0.8997
## P-Value [Acc > NIR] : 0.5484
##
## Kappa : 0
##
## Mcnemar's Test P-Value : 1.192e-07
##
## Sensitivity : 1.0000
## Specificity : 0.0000
## Pos Pred Value : 0.8997
## Neg Pred Value : NaN
## Prevalence : 0.8997
## Detection Rate : 0.8997
## Detection Prevalence : 1.0000
## Balanced Accuracy : 0.5000
##
## 'Positive' Class : 0
## # Variable importance
importance(rf_model)## 0 1 MeanDecreaseAccuracy MeanDecreaseGini
## gender -0.8969943 0.3657567 -0.7292315 3.612849
## age 1.7085166 -2.5981278 0.8340336 25.489796
## hypertension 0.1844302 -2.4982162 -0.5920865 4.547966
## heart_disease 1.0357498 1.1367636 1.4407012 4.280662
## ever_married 2.1435691 3.3105960 2.8879147 4.342828
## work_type 1.0682102 -0.6410287 0.8051711 12.293969
## Residence_type -0.9240259 -2.6602549 -1.7478691 3.708690
## avg_glucose_level 0.7972631 1.0714482 1.0651290 30.404742
## bmi 0.8993115 -0.9651927 0.5422147 27.415022
## smoking_status -1.3267743 -2.5081543 -2.1403343 9.547092varImpPlot(rf_model)
# Task Four: Deploy the prediction model
# Save the model
saveRDS(rf_model, "stroke_model.rds")
# Create a Plumber API to deploy the model
# Create a new R script named "plumber.R" and add the following:
# plumber.R
#' @post /predict
#' @param gender
#' @param age
#' @param hypertension
#' @param heart_disease
#' @param ever_married
#' @param work_type
#' @param Residence_type
#' @param avg_glucose_level
#' @param bmi
#' @param smoking_status
#' @response 200
function(gender, age, hypertension, heart_disease, ever_married, work_type, Residence_type, avg_glucose_level, bmi, smoking_status) {
# Load the saved model
model <- readRDS("stroke_model.rds")
# Create a new data frame with the inputs
new_data <- data.frame(
gender = factor(gender, levels = c("Male", "Female")),
age = as.numeric(age),
hypertension = as.numeric(hypertension),
heart_disease = as.numeric(heart_disease),
ever_married = factor(ever_married, levels = c("Yes", "No")),
work_type = factor(work_type, levels = c("Private", "Self-employed", "Govt_job", "Children", "Never_worked")),
Residence_type = factor(Residence_type, levels = c("Urban", "Rural")),
avg_glucose_level = as.numeric(avg_glucose_level),
bmi = as.numeric(bmi),
smoking_status = factor(smoking_status, levels = c("formerly smoked", "never smoked", "smokes", "Unknown"))
)
# Predict using the model
prediction <- predict(model, new_data)
return(as.character(prediction))
}## function(gender, age, hypertension, heart_disease, ever_married, work_type, Residence_type, avg_glucose_level, bmi, smoking_status) {
## # Load the saved model
## model <- readRDS("stroke_model.rds")
##
## # Create a new data frame with the inputs
## new_data <- data.frame(
## gender = factor(gender, levels = c("Male", "Female")),
## age = as.numeric(age),
## hypertension = as.numeric(hypertension),
## heart_disease = as.numeric(heart_disease),
## ever_married = factor(ever_married, levels = c("Yes", "No")),
## work_type = factor(work_type, levels = c("Private", "Self-employed", "Govt_job", "Children", "Never_worked")),
## Residence_type = factor(Residence_type, levels = c("Urban", "Rural")),
## avg_glucose_level = as.numeric(avg_glucose_level),
## bmi = as.numeric(bmi),
## smoking_status = factor(smoking_status, levels = c("formerly smoked", "never smoked", "smokes", "Unknown"))
## )
##
## # Predict using the model
## prediction <- predict(model, new_data)
##
## return(as.character(prediction))
## }Final Task: Findings and Conclusions
library(plumber) r <- plumb(“plumber.R”) r$run(port=8000)
library(packrat) library(rsconnect)
rsconnect::setAccountInfo(name=‘joih3r-wan0mohamad0hanis0bin0wan0hassan’, token=‘19A258BBD42A2E70A90A25E401535568’, secret=‘54JVgjDzkSsXl71JpSL1zYXeEofXEGQXC5S+Nwlu’) rsconnect::deployApp(‘path/to/your/app_or_project’)