prediction assignment.Rmd

---
title: "Prediction assignment"
author: "Milou Sep"
date: "2/29/2020"
output: html_document
---

Load testing and training data, and the caret package
```{r setup}
library(caret)
read.csv("pml-testing.csv")->testing
read.csv("pml-training.csv")->training
```

variables with missing values in test set are removed from test & training sets
```{r}
which(colSums(is.na(testing))==20)->missing.te
testing[,-missing.te]->test.cc

# which(colSums(is.na(training))==19622)->missing.tr
training[,-missing.te]->train.cc

```


Change all variables to numeric in both datasets
```{r}
train.cc <- data.frame(lapply(train.cc, function(x) as.numeric(x)))
 train.cc$classe<-as.factor(train.cc$classe)
 test.cc <- data.frame(lapply(test.cc, function(x) as.numeric((x))))

 #classe is niet aanwezig in testing?
 # testing$classe<-as.factor(testing$classe)
```

set-up 5-fold crossvalidation & train a random forest model.
All variables in data were used as predictors (msising values were ignored).
```{r}
set.seed(1234)
trainControl(method = 'cv', number=5)->control.cv
train(classe~., 
      method='rf', 
      data=train.cc, 
      trControl = control.cv)->rf.m
 #saveRDS(rf.m,'rfm.RDS')
```

Inspect the final model
```{r}
# readRDS('rfm.RDS')->rf.m
rf.m$finalModel
```
The OOB estimate of error rate is 0.01%

Predict the scores in the test data
```{r}
predict(rf.m,test.cc)->pred.rf
```

All new cases in class1
```{r}
plot(pred.rf)
```