slides.Rmd

---
title: "KiWi-Search"
subtitle: "House Price Comparator"
author: "Dziedziul ⚔ Salvini"
institute: "Universita Cattolicà Del Sacro Cuore"
date: "2019/11/18 (updated: `r Sys.Date()`)"
output:
  xaringan::moon_reader:
    css: [default, hygge , middlebury-fonts]
    lib_dir: libs
    nature:
      highlightStyle: github
      highlightLines: true
      countIncrementalSlides: false
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 300px
background-position: 50% 60%

# KiWi Logo

### Designed by Consumers for Consumers 
```{r include=FALSE}
libs = c("rvest", "magrittr", "stringr", "httr", "furrr", "plotly", "ggplot2", "DT", "readxl", "dplyr")
lapply(libs, require, character.only = TRUE)
```
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%
  
# Accomodation in Milan is a Mess
##Socio-Economic Circumstances:
- Too much *Demand* wrt to the available *Offer* 

--

- Rent Prices, for the same reason, are .red[**Overshooted**]
  
--
  
- .red[**Condominium**] is a really tough component of the monthly budget and actually It does not make so much sense

--
  
- Asymmetry of information between you and the agency Advisor

--
  
- Time factor: you are in a Hurry, houseolds are **Not**
  
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%
  
# The questions you come up 
##The most of us:
  
- Where do phisically I have to search?  --> **Immobiliare.it**<sup>[1]</sup> 
  
--
  
- Where do I need my apt to be? --> **select zone** within *Circonvalla* let's say
  
--
  
- Once you have spotted it '(if you made it') you should verify *if it is a good deal*
  
--

  
- Are they trying to steal my money? Am I going to pay a fair price?
  
--
  
- Hey, **I REALLY NEED SOMEONE TO TELL ME IF IT IS GOOD OR NOT**, let's find something which is similar to my option, oh I would say a *Comparable*

--

- Do you know how many are the chances to find a **Comparable** ?

.footnote[
[1] Here it is [immobiliare](https://www.immobiliare.it/), n1 player in online mrkt

]

---
background-image: url(https://video-images.vice.com/_uncategorized/1498664480808-Screen-Shot-2017-06-28-at-164017.png)
background-size: contain
---
# Odds

Now let's try this:
`TRY` to compute the probability to find an apartement given some certain characteristics (even location) equal to someone else with the **SAME** chars

- Randomly select 1 obs from the dataset

--

- Filter the dataset given the characteristics *(rand selct)* both .red[*Streched*] and .blue[*Strict*]

--

- Compare the `len` output of the `vec`  wrt to dataset `len`


---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# Make a **Comparison**!

```{r include=FALSE}
dataset = read_excel('dataset finale.xlsx') 
attach(dataset)
```

```{r eval=require('readxl'), echo=FALSE, warning=FALSE, tidy=FALSE}
dataset$tpprop = as.factor(dataset$tpprop)
dataset$status = as.factor(dataset$status)
dataset$heating = as.factor(dataset$heating)
dataset$ac = as.factor(dataset$ac)
dataset$enclass = as.factor(dataset$enclass)
dataset$contr = as.factor(dataset$contr)
dataset$furnished = as.factor(dataset$furnished)
dataset$kitchen = as.factor(dataset$kitchen)
dataset$nroom = as.factor(dataset$nroom)
dataset$lat = sub("(.{2})(*)", "\\1.\\2", dataset$lat) %>%  as.numeric()
dataset$long = sub("(.{1})(*)", "\\1.\\2", dataset$long) %>%  as.numeric()
```


## Here we pick a random row from the dataset:

```{r}
set.seed(123)
index = sample(nrow(dataset),1)
Pick1Sample = dataset[index,]
Pick1Sample[,1:6] %>% 
  knitr::kable('html')
  
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# Make a **Comparison** 2

.pull-left[

If you just .blue[do not really care] about the 
Location, and all of the conforts then: 

```{r highlight.output=c(4,5,6,7)}
SimpleQuery = dataset %>%
 filter(price >= 2400L & price <= 2600L) %>%
 filter(nroom %in% c("2","3")) %>%
 filter(sqmeter >= 90L & sqmeter <= 110L) %>%
 filter(condom >= 350L & condom <= 650L)
SimpleQuery[,1:5]
```

```{r highlight.output=c(1), echo=FALSE}
dim(SimpleQuery)[1]/dim(dataset)[1]
```

]
.pull-right[

But If you .red[**DO CARE**]


```{r highlight.output=c(4)}
SimpleQuery2 = dataset %>%
 filter(price >= 2400L & price <= 2600L) %>%
 filter(nroom %in% c("2","3")) %>%
 filter(sqmeter >= 90L & sqmeter <= 110L) %>%
 filter(condom >= 350L & condom <= 650L) %>% 
 filter(lat>= 45.4720	 & lat <= 45.4810	 ) %>% 
 filter(long>= 9.17650 & long <= 9.17960)
SimpleQuery2[,1:5]

```

*Our random obs... ^^*
]
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# So ...

1. you .red[*can't*] compare the offer you see in front of you with others

1. you .red[*can't*] really verify the house  -  trust the real estate agent

1. you .red[*can't*] even each time grab a train-plane to check



# .center[Now...]
--

# .center[**Am I going to pay a .red[Fair] price for an apartment I plan to rent?**]
---
background-image: url(https://media0.giphy.com/media/3o6Mbj2w67HnPQKgcE/giphy.gif)
background-size: cover


---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# Why don't we just...

1. Elaborate a tool that given some chars, computes the *estimated price* for that House

--
  
1. Build an application that is easy to understand and to use 

--
  
1. That can Give me a general overview of how distant my apt is from my .red[*Core Business*]
  
--
  
1. ... Also fancy, why not?
  
---
background-image: url(https://media.giphy.com/media/5z0cCCGooBQUtejM4v/giphy.gif)
background-size: contain

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%
  
# To Do ⚔ Doing
  
.pull-left[
1. Get data, **ETL**;
    
1. Preliminary **EDA**;
    
1. Create an online application;
    
1. Elaborate the **Model**;
    
1. Slides to present the work;
    
  ]

.pull-right[
1. **Scraping**<sup>[1]</sup> functions;
  
1. **ggplot2** & **Plotly** EDA;
  
1. Build [Shiny](https://rstudio.com/products/shiny/) application;
  
1. [H20](https://spark.rstudio.com/guides/h2o/) auto ML;
  
1. **xaringan**<sup>[2]</sup> slideshow to make it fancier!;
  ]

.footnote[
  [1] See the whole [Source Code](https://github.com/NiccoloSalvini/Web-Scraping/blob/master/Progetto%20di%20Computational.R) for the scraping 
  
  [2] See [xaringan](https://github.com/yihui/xaringan/) references
  ]

---
background-image: url(https://i.makeagif.com/media/6-20-2016/Si-tXL.gif)
background-size: 400px

# WAIT, but first why KiWi? 

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%
  
# .green[KiWi] bird 
  
- .green[**KiWi**] or kiwis are flightless birds native to New Zealand, in the genus Apteryx and family Apterygidae. Approximately the size of a domestic chicken, kiwi are by far the smallest living ratites (which also consist of ostriches, emus, rheas, and cassowaries)..

--


- The .green[**KiWi**] is unfortunately an animal at risk of *extinction<sup>[3]</sup>* due to hunting and also due to the introduction in New Zealand of the opossum, an animal that has become a predator, attacking the adults and devouring the eggs. To save the kiwi this animal is now bred in captivity.

--

- When the .green[**KiWi**] <sup>[2]</sup> female lays eggs something really amazing happens the female it is able to generate only one egg at a time of large dimensions which can almost be equal to the same adult animal size. But then the male, not the female, hatches the eggs.



.footnote[
  [1] See [here](https://www.greenme.it/informarsi/animali/kiwi-animale/) to find some other info 
  
  [2] Check [KiWi](https://it.wikipedia.org/wiki/Apteryx) for more references
  
  [3] Donate @[Here](https://www.theguardian.com/world/2018/jun/01/save-kiwi-new-zealand-protect-bird-species) to support the specie
  ]


---
background-image: url(https://assets3.thrillist.com/v1/image/2624055/size/tmg-article_tall.jpg)
background-size: cover

---
background-image: url(https://www.contiki.com/six-two/wp-content/uploads/2019/06/0208NWZD2018-1.jpg)
background-size: cover

---
background-image: url(https://media2.giphy.com/media/QmM2VFfudLjoc/giphy.gif)
background-size: cover

---
  
  
# So what KiWi bird and Apt research in Milan have in common?
- They are both into an **Extinction Crisis**

--
  
- Somewhat it seems you are the most interested part into your relationship (*kiwi father* & *kiwi mother* **against** *household/agency* & *tenants*)

--
  
- When .green[**KiWi**] gets into a relationship it stays there for 20 years. Yep, that is what happens when you signs a 4+4 rental
---
class: inverse, middle, center

# Scraping Functions
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%
  
#The very **beginning** part 1
  
- It all starts from a very basic URL, let's call it *Category*
```{r tidy=TRUE}
url  = "https://www.immobiliare.it/affitto-case/milano/?criterio=rilevanza&localiMinimo=1&localiMassimo=5&idMZona[]=10046&idMZona[]=10047&idMZona[]=10053&idMZona[]=10054&idMZona[]=10057&idMZona[]=10059&idMZona[]=10050&idMZona[]=10049&idMZona[]=10056&idMZona[]=10055&idMZona[]=10061&idMZona[]=10060&idMZona[]=10070&idMZona[]=10318&idMZona[]=10296&idMZona[]=10069"

```

--

- Then you generate all the remaming links starting from the **first** (*URL*) with the `str_c` function adding a piece of string in the last part, It stops at the 110<sup>th</sup> page
```{r tidy=FALSE}
list.of.pages.imm = c(url, str_c(url, '&pag=', 2:110))
list.of.pages.imm[2:8] %>% str_sub(-10,-1)
```

--

- From that links we are going to pick just some info
 1. **Text**
 1. **Price**
 1. **Rooms**
 1. **Sq Meters**
 1. **Primary Key**

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# The very **beginning** part 2

- I could not scrape the total of the data I need, basically for *2* main *reasons*:
 2. *Sometimes they are not present*
 2. *They are not all the data that I am looking for*
 
 ![McFitti](https://media0.giphy.com/avatars/TEAMFITTI/DGFe5RBetFM6.gif)
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# **Scrape** the price!
```{r highlight.output=c(1)}
scrapeprice.imm = function(url){
  
    x = GET(url, add_headers("user-agent" = "Gov employment data 
                                  scraper ([[niccopos@hotmail.it])"))
    web = read_html(x) %>% 
      html_nodes(css = ".lif__item:nth-child(1)") %>% 
      html_text() %>%
      str_replace_all("€","") %>% 
      str_replace_all("\\.","") %>% 
      str_trim() %>%  
      str_extract( "\\-*\\d+\\.*\\d*") %>%  
      str_replace_na() %>% 
      str_replace_all("NA", "Prezzo Su Richiesta")
    return(web)
}
```

This is the [URL](https://www.immobiliare.it/affitto-case/milano/?criterio=rilevanza&localiMinimo=1&localiMassimo=5&idMZona[]=10046&idMZona[]=10047&idMZona[]=10053&idMZona[]=10054&idMZona[]=10057&idMZona[]=10059&idMZona[]=10050&idMZona[]=10049&idMZona[]=10056&idMZona[]=10055&idMZona[]=10061&idMZona[]=10060&idMZona[]=10070&idMZona[]=10318&idMZona[]=10296&idMZona[]=10069), verify the prices 

```{r echo=FALSE, warning=FALSE}
plan(multisession)
some_urls = list.of.pages.imm[1:2] %>%  furrr::future_map(scrapeprice.imm) %>%  unlist
head(some_urls,25) 
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# **Scrape** the rooms!
```{r tidy=FALSE, eval=require('furrr'), highlight.output=c(1,4,9)}
scraperooms.imm = function(url){
    
    x <- GET(url, add_headers("user-agent" = "Gov employment data scraper ([[niccopos@hotmail.it])"))
    web = read_html(x) %>% 
      html_nodes(css = ".lif__item:nth-child(2) .text-bold") %>% 
      html_text() %>% 
      str_trim() %>% 
      as.numeric()
    
    return(web)
  }
```

```{r echo=FALSE, warning=FALSE}
plan(multisession)
some_others = list.of.pages.imm[1:2] %>%  furrr::future_map(scraperooms.imm) %>%  unlist
head(some_others,25) 
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# Here it goes for all the other *Functions* 
- for all the other data you need [**slide: the very beginning part 1**]

--

- Then you need a function that **aggregrates** the *sub-functions* you defined before.

```{r tidy=FALSE}
 get.data.caturl = function(urls){
    ad     = scrapetext.imm(urls)
    price  = scrapeprice.imm(urls)
    room   = scraperooms.imm(urls)
    sqmeter  = scrapespace.imm(urls)
    primar = scrapeprimarykey.imm(urls)
    
    combine = tibble(adtext  = ad,
                     monthlyprice   = price, 
                     nroom    = room,
                     sqmeter = sqmeter,
                     primary = primar)
    combine %>%
      select(primary, sqmeter, nroom, adtext, monthlyprice, vetrina)
    return(combine) 
 }
 
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# The **Advanced** part 1 

- Now you are looking for information inside the single apt address, here they are some from the **1**<sup>st</sup> page of immobiliare, if you do not trust us you can **verify it**, it is *real-time* 

```{r echo=FALSE, warning=FALSE}
scrapehref.imm = function(url){
    
    x <- GET(url, add_headers('user-agent' = 'Gov employment data scraper ([[niccopos@hotmail.it])'))
    web = read_html(x) %>% 
        html_nodes(css = '.text-primary a') %>% 
        html_attr('href') %>%
        as.character()
    return(web)
}
```



```{r highlight.output=c(1,4,9), echo=FALSE}
plan(multisession)
singles = list.of.pages.imm[1] %>% furrr::future_map(scrapehref.imm) %>% unlist 
singles[1:15]
```
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 8%

# The **Advanced** part 2 

- Now you get the access to be in the single link apt, that pretty looks like that: 

--

```{r}
browseURL(url = url)
```

.footnote[
[1] Check [here](https://www.immobiliare.it/annunci/77198460/) to single link for the house 
]
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# The **Advanced** part 2 
- For all the information you have seen there a specific function that picks up that
 1. **Latitude**                --> scrapelat.imm(html) 
 1. **Longitude**               --> scrapelong.imm(html)
 1. **Condominium cost**        --> scrapecondom.imm(html)
 1. **The Floor**               --> scrapefloor.imm(html)
 1. **The Age of the building** --> scrapeage.imm(html)
 1. *...*
 
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# Here it is one of them

- *Header* spoofing 
- *CSS* query of the whole page
- *Looping* into the page searching for "piano"
- Then if it present `paste0` the element, otherwise set the *NA*


```{r highlight.output= c(1,4,9)}
scrapefloor.imm = function(singolourl){
  Sys.sleep(0.4)
      x = GET(singolourl, add_headers('user-agent' = 'Gov employment data scraper ([[niccopos@hotmail.it])')) #<<
      web = read_html(x) %>%
        html_nodes(css ='.section-data .col-xs-12 .col-xs-12') %>% #<<
        html_text() %>%
        str_trim()
      m   = vector()
      for(i in 1:length(web)){ #<<
        if(web[i]=="Piano"){
          m[1] = paste0(web[i+1])} 
        else {m[1] == NULL}
      }
      m[is_empty(m)] = NA
      return(m)}
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# How it behaves?

```{r highlight.output=c(1,4,9)}
floors = singles %>% furrr::future_map(scrapefloor.imm) %>%  unlist
head(floors, n=15)
```
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# ... And that Goes on for all the others

Now, once again, build the aggregating function for the *second* type obs.

--

# **DO NOT SCARE YOURSLEF**

- It is somewhat you have *already seen* previuos 

--

- It *aggregates* all the different function defined
 
--
 
- All the functions pretty look like the same except for small changes (".red[piano]", ".red[ac]", ".red[heating system]")
---
```{r tidy=TRUE}
get.data.catsing = function(html){
  id   = scrapehouse.ID(html)
  lat  = scrapelat.imm(html)
  long = scrapelong.imm(html)
  location = take.location(html)
  condom = scrapecondom.imm(html)
  buildage = scrapeagebuild.imm(html)
  floor = scrapefloor.imm(html)
  indivsapt = scrapetype.imm(html)
  locali= scrapecompart.imm(html)
  tpprop = scrapeproptype.imm(html)
  status = scrapestatus.imm(html)
  heating = scrapeheating.imm(html)
  ac = scrapeaircondit.imm(html)
  date = scrapeaddate.imm(html)
  aptchar = scrapeaptchar.imm(html)
  photosnum = scrapephotosnum.imm(html)
  age = scrapeage.imm(html)
  enclass = scrapeenclass.imm(html)
  contr = scrapecontr.imm(html)
  combine = tibble(ID        = id,
                   LAT       = lat, 
                   LONG      = long,
                   LOCATION  = location,
                   CONDOM    = condom,
                   BUILDAGE  = buildage,
                   FLOOR     = floor,
                   INDIVSAPT = indivsapt,
                   LOCALI    = locali,
                   TPPROP    = tpprop,
                   STATUS    = status,
                   HEATING   = heating,
                   AC        = ac,
                   DATE      = date,
                   APTCHAR   = aptchar,
                   PHOTOSNUM = photosnum,
                   AGE       = age,
                   ENCLASS   = enclass,
                   CONTR     = contr)
  combine %>% 
    select(ID, LAT, LONG, LOCATION, CONDOM, FLOOR, INDIVSAPT, LOCALI, TPPROP, STATUS, HEATING,
           AC, DATE, APTCHAR, PHOTOSNUM, AGE, ENCLASS, BUILDAGE, CONTR)
  return(combine) 
}
```

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# Why did you this in that way?
## 2 main *factors*:

- Looping into the page was a way to undergo to the issues of missing values, the CSS query can be very precise but sometimes information are placed in the wrong order
- When you unlist a   `vec`   if there is inside a   `NULL`   then the vect is crunched and you loose some dimensionality, so I set   `NULL`   values equal to   `NA`   in order to have same lenght vectors
---
# Then datasets become two!


--


1. The one scraped from the category    **SLIDE** : 39
1. The one scraped from the single link **SLIDE** : 44



![datasets](https://i.stack.imgur.com/Zj9zP.png)


---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# Done

data is available in a MySql database.
I have it also in local:


```{r echo=FALSE}
dataset = read_excel('dataset finale.xlsx') 
DT::datatable(
  head(dataset[,1:5], 10),
  fillContainer = FALSE, options = list(pageLength = 8)
)

```

```{r data wrang, eval=require('readxl'), echo=FALSE, warning=FALSE, tidy=FALSE}
dataset$tpprop = as.factor(dataset$tpprop)
dataset$status = as.factor(dataset$status)
dataset$heating = as.factor(dataset$heating)
dataset$ac = as.factor(dataset$ac)
dataset$enclass = as.factor(dataset$enclass)
dataset$contr = as.factor(dataset$contr)
dataset$furnished = as.factor(dataset$furnished)
dataset$kitchen = as.factor(dataset$kitchen)
dataset$nroom = as.factor(dataset$nroom)
dataset$lat = sub("(.{2})(*)", "\\1.\\2", dataset$lat) %>%  as.numeric()
dataset$long = sub("(.{1})(*)", "\\1.\\2", dataset$long) %>%  as.numeric()
```


---
class: inverse, middle, center

# EDA part
---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%

# How it is composed 
```{r tidy= FALSE, eval= require('DataExplorer')}
DataExplorer::introduce(dataset)
```
--

here it is a [report](report.html) of the composition of the dataset, *Let's take a look*

.pull-left[
- Basic Statistics;
 - Raw Counts;
 - Percentages;
- Data Structure;
- Missing Data Profile;
    
    ]

.pull-right[
- Univariate Distribution;
 - Histogram;
 - Bar Chart (by frequency);
 - QQ Plot;
- Correlation Analysis;
- Principal Component Analysis
  ]

---
background-image: url(https://i.pinimg.com/originals/1b/db/46/1bdb46cf377ae1c06617d6b9bfc54793.png)
background-size: 100px
background-position: 90% 6%
# Map, blue points are **Houses**

```{r out.width='100%', fig.height=7, eval=require('leaflet'), warning= FALSE, echo=FALSE}
leaflet() %>%
  setView(lng = 9.18994, lat = 45.4703, zoom = 13) %>% 
  addProviderTiles(providers$OpenStreetMap.Mapnik) %>%
  addCircles(dataset$long, dataset$lat, label = paste0('this is the price', as.character(dataset$price)))
```
---
# Some Other `Explorations`

```{r out.width='100%', fig.height=7, warning= FALSE, echo=FALSE, fig.retina= 3}
ddataset = dataset %>%
 filter(dataset$price >= 500L & dataset$price <= 15128L)
m = ggplot(ddataset) +
 aes(x = ddataset$price, fill = ddataset$nroom) +
 geom_density(adjust = 1L) +
 scale_fill_hue() +
 labs(x = "price", y = "nrooms", title = "Density Plot", subtitle = "price VS nrooms")
ggplotly(m)
```


---
# Floor `investigation`

```{r out.width='100%', fig.height=7, warning= FALSE, echo=FALSE, fig.retina= 3}

floordataset = dataset %>%
  filter(price >= 500L & price <= 10581L) %>%
  filter(!(floor %in% c("NA", "12", "10")))

p <- ggplot(data = floordataset,aes(x=floordataset$nroom, y=floordataset$price)) +
  geom_boxplot(fill=NA, alpha=0.5) +
  geom_jitter(aes(colour=floordataset$floor, text=paste("price is good")), width=0.25, alpha=0.5) +
  geom_hline(yintercept=mean(floordataset$price)) +
  labs(title = "How much the floors affect the price given the number of rooms",
       x = "nroom",
       y = "price")
ggplotly(p)
```

---
# `Facet Plot` Condominium

```{r out.width='100%', fig.height=7, warning= FALSE, echo=FALSE, fig.retina= 3}
ldataset = dataset %>%
 filter(price >= 500L & price <= 5047L) %>%
 filter(condom >= 0L & condom <= 1030L) %>%
 filter(!(floor %in% c("NA", "12", "10")))

q = ggplot(ldataset) +
 aes(x = price, y = condom, colour = condom, size = condom) +
 geom_point() +
 scale_color_gradient() +
 theme_minimal() +
 facet_wrap(vars(nroom), scales = "free_x")
ggplotly(q)
```


---
# **Narrow** Boxplot 


```{r out.width='100%', fig.height=7, warning= FALSE, echo=FALSE, fig.retina= 3}
vdataset = dataset %>%
 filter(condom >= 200L & condom <= 500L) %>%
 filter(!(heating %in% "Unknown")) %>%
 filter(!(ac %in% c("Assente", "NA", "Assente, solo freddo", "Predisposizione impianto, solo freddo", 
    "Predisposizione impianto", "Presente", "Assente, solo caldo")))

r = ggplot(vdataset) +
 aes(x = nroom, y = condom, fill = nroom) +
 geom_boxplot() +
 scale_fill_hue() +
 theme_minimal() +
 facet_wrap(vars(floor), scales = "free")
ggplotly(r)
```

---
# Heating **vs** Condominium
```{r out.width='100%', fig.height=7, warning= FALSE, echo=FALSE, fig.retina= 3}
s = ggplot(vdataset) +
 aes(x = heating, y = condom, fill = heating) +
 geom_boxplot() +
 scale_fill_hue() +
 theme_minimal()
ggplotly(s)
```

---
# **License**



.left[
<a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/"><img alt="Licenza Creative Commons" style="border-width:0" src="https://i.creativecommons.org/l/by-sa/4.0/88x31.png" /></a><br />Quest'opera è distribuita con Licenza <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribuzione - Condividi allo stesso modo 4.0 Internazionale</a>. ]