test.Rmd

```{r}
df <- read.csv("./data/raw/framingham.csv")

df
```

```{r}
sampl <- sapply(df, class)
sampl
```

```{r}
library(psych)

describe(df)
```

```{r}
library(purrr)

encode_ordinal <- function(x, order = unique(x)) {
  x <- as.numeric(factor(x, levels = order, exclude = NULL))
  x
}

colTypes <- map(df, class)

for (col in colnames(df)) {
  if (colTypes[col] == 'character') {
    df[[col]] <- encode_ordinal(df[[col]])
  }
}

df
```

```{r}
df <- df[complete.cases(df),]
df
```

```{r}
as.factor(df$'TenYearCHD')
```

```{r}
library(caret)

df$'TenYearCHD' <- as.factor(df$'TenYearCHD')

inTrain <- createDataPartition(as.factor('TenYearCHD'), times = 1, p = .75, list = F)
data_train <- df[inTrain, ]
data_test <- df[-inTrain, ]
```

```{r}
logit <- train(as.factor(TenYearCHD)~., data = df, method = 'ranger', metric='Accuracy')
logit
```
```{r}
data_test$pred <- predict(logit, data_test)

data_test$factor_pred <- as.factor(data_test$pred)
data_test$factor_truth <- as.factor(data_test$TenYearCHD)

precision <- posPredValue(data_test$factor_truth, data_test$factor_pred)
recall <- sensitivity(data_test$factor_truth, data_test$factor_pred)

cm <- confusionMatrix(data_test$pred, data_test$TenYearCHD)

accuracy <- cm$overall[1]
accuracy

cm
```

```{r}
predict <- predict(logit, data_test, type = 'response')
# confusion matrix
table_mat <- table(data_test$TenYearCHD, predict > 0.5)
table_mat
```

```{r}
accuracy_Test <- sum(diag(table_mat)) / sum(table_mat)
accuracy_Test
```