---
title: "predictive_model"
output: pdf_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

# Data
```{r}
library(caret)
library(pROC)
library(daDoctoR)
library(dplyr)
```

Import
```{r}
trainData<-read.csv("/Users/au301842/PhysicalActivityandStrokeOutcome/data/trainData.csv",)
testData<-read.csv("/Users/au301842/PhysicalActivityandStrokeOutcome/data/testData.csv",)
```


# Prediction
Inspiration: https://stackoverflow.com/questions/30366143/how-to-compute-roc-and-auc-under-roc-after-training-using-caret-in-r and https://www.machinelearningplus.com/machine-learning/caret-package/

## Early visualisation

```{r}
featurePlot(x = trainData %>% select(!matches("pase_drop")),
            y = factor(trainData$pase_drop), 
            plot = "box",
            strip=strip.custom(par.strip.text=list(cex=.7)),
            scales = list(x = list(relation="free"), 
                          y = list(relation="free")))

featurePlot(x = trainData %>% select(!matches("pase_drop")),
            y = factor(trainData$pase_drop), 
            plot = "density",
            strip=strip.custom(par.strip.text=list(cex=.7)),
            scales = list(x = list(relation="free"), 
                          y = list(relation="free")))
```


```{r}
subsets <- c(1:10, 15, 18,33)

ctrl <- rfeControl(functions = rfFuncs,
                   method = "repeatedcv",
                   repeats = 5,
                   verbose = FALSE)

lmProfile <- rfe(x = trainData %>% select(!matches("pase_drop")), 
                 y = trainData$pase_drop,
                 sizes = subsets,
                 rfeControl = ctrl)

lmProfile
```


```{r}
set.seed(1000)

forest.model <- train(pase_drop ~., trainData)

result.predicted.prob <- predict(forest.model, testData, type="prob") # Prediction

result.roc <- roc(factor(testData$pase_drop), result.predicted.prob$no) # Draw ROC curve. 

plot(result.roc, print.thres="best", print.thres.best.method="closest.topleft")

result.coords <- coords(result.roc, "best", best.method="closest.topleft", ret=c("threshold", "accuracy"))
print(result.coords)#to get threshold and accuracy
```

```{r}
library(MLeval)

myTrainingControl <- trainControl(method = "cv", 
                                  number = 10, 
                                  savePredictions = TRUE, 
                                  classProbs = TRUE, 
                                  verboseIter = TRUE)

randomForestFit = train(x = trainData[,1:32], 
                        y = as.factor(trainData$pase_drop), 
                        method = "rf", 
                        trControl = myTrainingControl, 
                        preProcess = c("center","scale"), 
                        ntree = 50)

x <- evalm(randomForestFit)

x$roc

x$stdres
```