Ejercicio_final_R.Rmd

```{r}
library(dplyr)
library(tidyr)
```

```{r}
df <- read.csv("Airbnb_Milan.csv")
```

```{r}
new_df <- subset(df, select = c("host_is_superhost", "host_identity_verified", "bathrooms", "bedrooms", "daily_price", "security_deposit", "minimum_nights", "number_of_reviews", "review_scores_rating"))
```

```{r}
new_df
```

```{r}
new_df$host_is_superhost <- recode(as.character(new_df$host_is_superhost),
                              "0" = "SI",
                              "1" = "NO") %>% as.factor
```

```{r}
new_df$host_identity_verified <- recode(as.character(new_df$host_identity_verified),
                                    "0" = "VERIFICA",
                                    "1" = "NO VERIFICA") %>% as.factor 
```

```{r}
summary(new_df)
```
                         
```{r}
new_df[new_df$minimum_nights <= 7,]
```
                          
```{r}
tapply(new_df$daily_price, new_df$host_identity_verified, mean)
```

```{r}
tapply(new_df$number_of_reviews, new_df$host_is_superhost, max)
```

```{r}
tapply(new_df$review_scores_rating, new_df$host_is_superhost, mean)
```

```{r}
categorize.score <- function(df.column){
  
  resultados <- c()
  i <- 1
  
  for (fila in 1:length(df.column)) {
    if (df.column[fila] <= 49){
      resultados[i] <- "NO ACONSEJABLE"
      i <- i + 1
      
    } else if ((df.column[fila] >= 50) && (df.column[fila] <= 75)) {
      resultados[i] <- "ESTÁNDAR"
      i <- i + 1
      
    } else {
      resultados[i] <- "TOP"
      i <- i + 1
    }
  }
  
  return (as.factor(resultados))
}
new_df$CATEGORIA <- categorize.score(new_df$review_scores_rating)
```

```{r}
table(new_df$CATEGORIA)
```

```{r}
library(ggplot2)
```

```{r}
ggplot(new_df) +
geom_histogram(mapping = aes(daily_price), color = "red", fill = "lightblue", alpha = 0.6)
```

```{r}
ggplot(new_df) + 
  geom_point(mapping = aes(x = bedrooms, y = bathrooms)) + geom_smooth(aes(x = bedrooms, y = bathrooms), method = lm)
```

```{r}
ggplot(new_df) +
geom_histogram(mapping = aes(number_of_reviews, fill = host_identity_verified), color = "red" )
```

```{r}
ggplot(new_df) +
  geom_density(mapping = aes(x = host_is_superhost, y = security_deposit, colour = CATEGORIA)) + facet_wrap(~CATEGORIA)
```