29-adults.Rmd

# Case study - The adults dataset.  

## Introduction
The adult data set is another famous one from the **UCI - machine learning repository**.  
The idea is to predict whether income exceeds $50K/yr based on census data. Also known as "Census Income" dataset.  Extraction was done by Barry Becker from the 1994 Census database.  

Load the libraries
```{r message=FALSE, warning=FALSE}
library(stringr)
```


## Import the data 
```{r message=FALSE, warning=FALSE}
df <- read_csv("dataset/adult.csv")
glimpse(df)
```

## Tidy the data
Let's check the level of missing data
```{r}
map_dbl(df, function(.x) sum(is.na(.x)))
```
No missing data!  That's great news. 

Before we change the <chr> variables into factors, let's see what type of levels we have.
```{r}
df %>% select_if(is.character) %>% map_if(is.character, unique)
```
Allright, so maybe there were no NA, but there are quite a few "?"

The "?" should probably be replaced with NAs.
```{r message=FALSE, warning=FALSE}
df <- read_csv("dataset/adult.csv", na = c("NA", "?"))

# Let's redo a check on the NA now
map_int(df, function(.x) sum(is.na(.x)))
```

Let's now rework the column names to better fit our naming conventions
```{r}
colnames(df) <- c("age", "working_class", "final_weight", "education", "education_num", "marital_status", 
                  "occupation", "relationship", "race", "gender", "capital_gain", "capital_loss", "hours_per_week", 
                  "native_country", "above_50k")
```


```{r}
df2 <- df %>% mutate_if(is_character, as.factor)

levels(df2$working_class)

summary(df2)
```