1 Variables.Rmd

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyr)
library(pander)
require(lattice)
require(ggplot2)
library(dplyr)
library(tidyverse)
source("Functions_TFM.R")
source("Functions.R")
```

# Importamos el dataset

```{r}
tabla.raw<-read.csv("./Data/IdealistaDepurado.csv",sep=",",encoding="UTF-8")
head(tabla.raw)
```
# Duplicados

- Exactos

```{r}
dups_identicos <- tabla.raw[duplicated(tabla.raw[2:30]),]
dups_identicos
```

- Todo excepto localización

```{r}
dups <- tabla.raw %>% group_by_at(vars(-id,-lat,-lon,-calle,-exacta)) %>% dplyr::filter(n()>1) %>% ungroup() %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
dups
```

```{r}
dummy <- dplyr::select(tabla.raw,-id,-lat,-lon,-calle,-exacta)
tabla.raw[(duplicated(dummy)),] %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
```

Elimino primero los que tienen dirección exacta = 0:

```{r}
#tabla.raw[(duplicated(dummy) | duplicated(dummy, fromLast=TRUE)),]
to.delete <- tabla.raw[(duplicated(dummy) & tabla.raw$exacta==0),]
to.delete %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
to.delete.id <- tabla.raw[(duplicated(dummy) & (tabla.raw$exacta==0)),1]
```


```{r}
tabla2 <- tabla.raw[ ! tabla.raw$id %in% to.delete.id, ]

tabla2 %>% group_by_at(vars(-id,-lat,-lon,-calle,-exacta)) %>% dplyr::filter(n()>1) %>% ungroup() %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
```
Ordeno al revés y así elimino todos los que tienen exacta = 0:

```{r}
dummy2 <- dplyr::select(tabla2,-id,-lat,-lon,-calle,-exacta)
tabla2[(duplicated(dummy2, fromLast=TRUE)),] %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
to.delete.id2 <- tabla2[(duplicated(dummy2, fromLast=TRUE)),1]
tabla <- tabla2[ ! tabla2$id %in% to.delete.id2, ]
```
Ya no quedan duplicados:

```{r}
dummy3 <- dplyr::select(tabla,-id,-lat,-lon,-calle,-exacta)
tabla[(duplicated(dummy3, fromLast=TRUE)),] %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
tabla %>% group_by_at(vars(-id,-lat,-lon,-calle,-exacta)) %>% dplyr::filter(n()>1) %>% ungroup() %>% dplyr::select(id,price,exacta,calle) %>% arrange(desc(price))
```


# DISTANCIAS

```{r}
Descarga_OSM_points<-function(ciudad,key,value){
  q <- getbb(ciudad) %>%
      opq() %>%
       add_osm_feature(key, value)
  return(osmdata_sf(q))
}
mapa1 <- opq(bbox = "Madrid, Spain")
```

- Hospitales:

```{r}
Hospitales<-Descarga_OSM(ciudad="Madrid",key='amenity',value = "hospital")
```

- Centros Comerciales:

```{r}
CentrosComerciales<-Descarga_OSM(ciudad="Madrid, Spain",key='shop',value = "mall")
```

- Transporte Público:

```{r, fig.heigth=20}
#Descargo datos OSM
Metro<-Descarga_OSM_points(ciudad="Madrid", key='public_transport', value = "station")
#Extraigo coordenadas de los puntos
coords_metro <- do.call(rbind, st_geometry(Metro$osm_points)) %>% 
    as_tibble() %>% setNames(c("lon","lat"))
```

- Colegios:

```{r}
Colegios<-Descarga_OSM(ciudad="Madrid, Spain",key='amenity',value = "school")
```

- DISTANCIAS

```{r}
#CC
coordenadas<-as.data.frame(gCentroid(CentrosComerciales[[2]], byid=TRUE)@coords)
Distancias<-distm(cbind(tabla$lon,tabla$lat),cbind(coordenadas$x,coordenadas$y),fun = distCosine )/1000
tabla$dist_cc<-round(apply(Distancias,1,min),2)
tabla$dens_cc<-apply((Distancias<1)*1,1,sum)
```

```{r}
#Hospitales
coordenadas<-as.data.frame(gCentroid(Hospitales[[2]], byid=TRUE)@coords)
Distancias<-distm(cbind(tabla$lon,tabla$lat),cbind(coordenadas$x,coordenadas$y),fun = distCosine )/1000
tabla$dist_hospital<-round(apply(Distancias,1,min),4)
tabla$dens_hospital<-apply((Distancias<1)*1,1,sum)
```

```{r}
#Metro
#coordenadas<-as.data.frame(gCentroid(Hospitales[[2]], byid=TRUE)@coords)
Distancias<-distm(cbind(tabla$lon,tabla$lat),cbind(coords_metro$lon,coords_metro$lat),fun = distCosine )/1000
tabla$dist_tp<-round(apply(Distancias,1,min),4)
tabla$dens_tp<-apply((Distancias<1)*1,1,sum)
```

```{r}
#Colegios
coordenadas<-as.data.frame(gCentroid(Colegios[[2]], byid=TRUE)@coords)
Distancias<-distm(cbind(tabla$lon,tabla$lat),cbind(coordenadas$x,coordenadas$y),fun = distCosine )/1000
tabla$dist_colegios<-round(apply(Distancias,1,min),4)
tabla$dens_colegios<-apply((Distancias<1)*1,1,sum)
head(tabla)
```

```{r}
#Sol
Distancias<-distm(cbind(tabla$lon,tabla$lat),cbind(-3.703676643104174,40.41672517384556),fun = distCosine )/1000
tabla$dist_centro<-round(apply(Distancias,1,min),4)
head(tabla)
```

# Inconsistencias

```{r}
tabla_inconsistencias <- tabla 
tabla_inconsistencias <- tabla_inconsistencias %>% 
  mutate(inc_neg = ((metros<=0) | (habitaciones <0) | (price <0) | (baths <0)),
         inc_tipo = ((piso==0) & (atico ==0) & (duplex == 0) & (chalet ==0) & (estudio ==0)),
         inc_estado = ((obra.nueva==0) & (buen.estado ==0) & (reformar == 0)),
         inc_planta = planta>23,
         inc_dist = (dist_cc>5) | (dist_tp >4) | (dist_hospital >5) | (dist_colegios >4))
```

```{r}
tabla_inconsistencias %>%
  dplyr::filter(inc_neg | inc_tipo | inc_estado | inc_planta | inc_dist) %>%
  dplyr::select(id,piso,atico,duplex,chalet,planta,inc_neg, inc_planta,inc_tipo,inc_estado,inc_dist, dist_cc, dist_tp, dist_hospital, dist_colegios)
```

# RESUMEN
```{r}
summary(tabla)
```
```{r}
glimpse(tabla)
```

# Histogramas

- **Variables numéricas**

```{r include=FALSE}
Hist1 <- function(bbdd_fff,response,predicted,var,n,breaks=8){
  
  names<-colnames(bbdd_fff)
  bbdd_fff$predicted<-predicted
  bbdd_fff$response<-response
  bbdd_fff$VIVO<-1
  q1<- bbdd_fff %>% 
    group_by(cut(var,breaks=breaks)) %>% 
    summarise(Exposicion = sum(VIVO),
              Frecuencia = sum(response)/sum(VIVO),
              Predicted = sum(predicted)/sum(VIVO)
              ) 
 
  q1<-as.data.table(q1)
  
  c1<-q1[,1]
  q1$c1<-q1[,1]
  
  ff<-ggplot(q1,aes(x=c1)) + 
    geom_bar(aes(y=(Exposicion/sum(Exposicion))*mean(Frecuencia)*2, fill="% de Exposicion"), stat = "identity")+ 
    geom_point(aes(y=Frecuencia, colour="Frecuencia"), group=1)+
    geom_line(aes(y=Frecuencia, colour="Frecuencia"), group=1)+
    geom_point(aes(y=Predicted, colour="Predicted"), group=1)+
    geom_line(aes(y=Predicted, colour="Predicted"), group=1)+    
    xlab("") + ylab("") +
    #scale_y_continuous(limits = c(0, 0.45),breaks = c(0:45/100))+
    theme(legend.position = "none", axis.text.x = element_text(angle = 60)) +
    ggtitle(names[n])
  
  return(ff)
  
  
}

```


```{r warning=FALSE, message=FALSE}
#Defino vector de numericas
numericas_hist<-dplyr::select(tabla,price,lon,lat,precio.metro,planta,construccion,metros,habitaciones,baths, dist_cc, dens_cc, dist_hospital, dens_hospital, dist_tp, dens_tp, dist_colegios, dens_colegios, dist_centro)
#Represento
for (i in 1:ncol(numericas_hist)){
pr<-Hist1(numericas_hist,response = numericas_hist[,1],predicted = 0,var = numericas_hist[,i],n=i,breaks = 10)
plot(pr)
}
```

```{r}
#Vector de double
numericas_dbl<-dplyr::select(tabla,precio.metro, price,lon,lat,precio.metro,planta,construccion,metros, dist_cc, dens_cc, dist_hospital, dens_hospital, dist_tp, dens_tp, dist_colegios, dens_colegios)
#
for (i in names(numericas_dbl)){
hist(numericas_dbl[[i]], main=colnames(numericas_dbl[i]), xlab=colnames(numericas_dbl[i]))
}
```

- **Variables integer**

```{r}
#Vector de enteros
numericas_int<-dplyr::select(tabla,habitaciones, baths, terraza, ascensor, aire, exacta, garaje, piscina, zonas.verdes, trastero, armarios, buen.estado, reformar, exterior, balcon, chalet, atico, duplex, estudio, piso, planta, dens_cc, dens_hospital, dens_tp, dens_colegios)

#Represento
for (i in names(numericas_int)){
barplot(table(numericas_int[[i]]), main=colnames(numericas_int[i]), xlab=colnames(numericas_int[i]))
}
```


# Distritos

```{r}
bp<- ggplot(tabla, aes(x="", y=distrito, fill=distrito))+
  geom_bar( stat = "identity") +
  coord_polar("y", start=0)
bp
```

```{r}
ggplot(tabla, aes(x=reorder(distrito, distrito, function(x)-length(x)))) +
  geom_bar(fill="tomato")+
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))
```
# Correlaciones

Problema con chalet y Obra Nueva!

```{r fig.width=17, fig.heigth=17}
numericas<-dplyr::select(tabla,-calle,-distrito,-chalet, -obra.nueva,-id)
cr <- cor(numericas, use="complete.obs")
ggcorrplot(cr, hc.order = TRUE,type = "lower",lab = TRUE)
```
# Outliers

```{r}
#Vector de double
numericas_outliers<-dplyr::select(tabla,price, metros, precio.metro, planta, construccion, habitaciones, baths, dist_cc, dens_cc, dist_hospital, dens_hospital, dist_tp, dens_tp, dist_colegios, dens_colegios)
for (i in names(numericas_outliers)){
boxplot(numericas_outliers[[i]], main=colnames(numericas_outliers[i]), xlab=colnames(numericas_outliers[i]))
}
```

# Dispersión con precio

```{r}
for (i in names(numericas_int)){
plot(numericas_int[[i]], log(tabla$precio.metro), main=colnames(numericas_int[i]), xlab=colnames(numericas_int[i]))
}

```
```{r}
for (i in names(numericas_dbl)){
plot(numericas_dbl[[i]], log(tabla$precio.metro), main=colnames(numericas_dbl[i]), xlab=colnames(numericas_dbl[i]))
}

```

```{r}
glimpse(tabla)
```

# Transformaciones
```{r}
tabla <- tabla %>%
  mutate(log.pm=log(price/metros),
         log.dist_hospital=log(dist_hospital),
         log.dist_tp=log(dist_tp),
         log.dist_colegios=log(dist_colegios))
summary(tabla)
```

```{r}
boxplot(tabla$log.pm)
boxplot(tabla$log.dist_hospital)
boxplot(tabla$log.dist_tp)
boxplot(tabla$log.dist_colegios)
```


# Mapas

- Hospitales:

```{r}
leaflet(Hospitales[[1]]) %>% addTiles() %>% addPolygons(data = Hospitales[[2]], col = "red",label =Hospitales[[3]] ) %>% addCircles()
```
```{r}
#names(st_geometry(Hospoints$osm_points)) = NULL
#names(st_geometry(Hospoints$osm_lines)) = NULL
#names(st_geometry(Hospoints$osm_polygons)) = NULL
#leaflet(Hospoints$osm_polygons) %>% addTiles() %>% # carga el mosaico base de OSM 
#  addPolylines(color ="blue",label = ~Hospoints$osm_lines$name ) %>%
#  addPolygons(color="yellow",label = ~Hospoints$osm_polygons$name) %>% 
#  addCircleMarkers(data=Hospoints$osm_points,radius = 7, color = "red",
#                   popup= ~Hospoints$osm_points$name)# carga la capa de puntos mercadonas
```

- Centros Comerciales:

```{r}
leaflet(CentrosComerciales[[1]]) %>% addTiles() %>% addPolygons(data = CentrosComerciales[[2]], col = "red",label =CentrosComerciales[[3]] ) %>% addCircles()
```

```{r}
#CCpoints<-Descarga_OSM_points(ciudad="Madrid", key='shop', value = "mall")
```

```{r}
#names(st_geometry(CCpoints$osm_points)) = NULL
#names(st_geometry(CCpoints$osm_lines)) = NULL
#names(st_geometry(CCpoints$osm_polygons)) = NULL
#leaflet(CCpoints$osm_polygons) %>% addTiles() %>% # carga el mosaico base de OSM 
#  addPolylines(color ="blue",label = ~CCpoints$osm_lines$name ) %>%
#  addPolygons(color="yellow",label = ~CCpoints$osm_polygons$name) %>% 
#  addCircleMarkers(data=CCpoints$osm_points,radius = 7, color = "red",
#                   popup= ~CCpoints$osm_points$name)# carga la capa de puntos mercadonas
```

- Transporte Público:

```{r, fig.heigth=20}
#our background map
#mad_map <- get_map(getbb("Madrid"), maptype = "toner-background")
#final map
#ggmap(mad_map)+
#  geom_sf(data = Metro$osm_points,
#          inherit.aes = FALSE,
#          colour = "#238443",
#          fill = "#004529",
#          alpha = .5,
#          size = 4,
#          shape = 21)+
#  labs(x = "", y = "")
```


```{r}
names(st_geometry(Metro$osm_points)) = NULL
names(st_geometry(Metro$osm_lines)) = NULL
names(st_geometry(Metro$osm_polygons)) = NULL
leaflet(Metro$osm_polygons) %>% addTiles() %>% # carga el mosaico base de OSM 
  addPolylines(color ="blue",label = ~Metro$osm_lines$name ) %>%
  addPolygons(color="yellow",label = ~Metro$osm_polygons$name) %>% 
  addCircleMarkers(data=Metro$osm_points,radius = 7, color = "red",
                   popup= ~Metro$osm_points$name)# carga la capa de puntos mercadonas
```

- Colegios:

```{r, fig.heigth=20}
#Descargo datos OSM
#Colegio<-Descarga_OSM_points(ciudad="Madrid",key="amenity", value="school")
#Extraigo coordenadas de los puntos
#coords_colegio <- do.call(rbind, st_geometry(Colegio$osm_points)) %>% 
#    as_tibble() %>% setNames(c("lon","lat"))

#our background map
#mad_map <- get_map(getbb("Madrid"), maptype = "toner-background")
#final map
#ggmap(mad_map)+
#  geom_sf(data = Metro$osm_points,
#          inherit.aes = FALSE,
#          colour = "#238443",
#          fill = "#004529",
#          alpha = .5,
#          size = 4,
#          shape = 21)+
#  labs(x = "", y = "")
```

```{r}
#names(st_geometry(Colegio$osm_points)) = NULL
#names(st_geometry(Colegio$osm_lines)) = NULL
#names(st_geometry(Colegio$osm_polygons)) = NULL
#leaflet(Colegio$osm_polygons) %>% addTiles() %>% # carga el mosaico base de OSM 
#  addPolylines(color ="blue",label = ~Colegio$osm_lines$name ) %>%
#  addPolygons(color="yellow",label = ~Colegio$osm_polygons$name) %>% 
#  addCircleMarkers(data=Colegio$osm_points,radius = 7, color = "red",
#                   popup= ~Colegio$osm_points$name)# carga la capa de puntos mercadonas
```

```{r}
leaflet(Colegios[[1]]) %>% addTiles() %>% addPolygons(data = Colegios[[2]], col = "red",label =Colegios[[3]] ) %>% addCircles()
```