Bikes_in_Hamburg.Rmd

---
title: "first"
author: "Friedrich Preußer"
date: "2/10/2021"
output: html_document
---

```{r setup, include=FALSE}
library(dplyr)
library(ggplot2)
library(osmdata)
library(sf)
library(purrr)
library(here)
library(viridis)
```


#get location of Hamburg
```{r}
city <- "Hamburg"
getbb(city)

```

#get first overview of Hamburg
```{r}
streets <- city %>%
  opq() %>%
  add_osm_feature(key = "highway", 
                  value = c("motorway", "primary", 
                            "secondary")) %>%
  osmdata_sf()

#river
river <- city %>%
  opq() %>%
  add_osm_feature(key = "waterway", value = "river") %>%
  osmdata_sf()

ggplot() +
  geom_sf(data = streets$osm_lines,
          inherit.aes = FALSE,
          color = "black",
          size = .4,
          alpha = .8)

#smaller
ggplot() +
  geom_sf(data = streets$osm_lines,
          inherit.aes = FALSE,
          color = "black",
          size = .4,
          alpha = .8) +
     coord_sf(xlim = c(9.75, 10.2),
            ylim = c(53.5, 53.65),
            expand = FALSE)

```

#load data
```{r}

#list of stations
stations <- read.csv(file.path(here(), "data", "bikeshare_stations_hh.csv"))
stations_sf <- stations %>%
  st_as_sf(coords = c("lon", "lat"), crs=4326)

#list of trips
trips <- read.csv(file.path(here(), "data", "bikeshare_trips_hh.csv"))

#landuse in hamburg
landuse <- readRDS(file.path(here(), "data", "landuse_sf.rds"))

#fix different GDAL versions
#see here: https://stackoverflow.com/questions/61286108/error-in-cpl-transformx-crs-aoi-pipeline-reverse-ogrcreatecoordinatetrans
st_crs(landuse) <- 25832

landuse_sf <- landuse %>%
  st_transform(., 4326)
```

#plot landuse
```{r}

ggplot(landuse_sf) +
  geom_sf(aes(fill=fclass), lwd = 0) +
  them_void()

```

#plot stations and landuse
```{r}


ggplot() +
  geom_sf(data=landuse_sf, aes(fill=fclass), lwd = 0) +
  geom_sf(data = streets$osm_lines,
        inherit.aes = FALSE,
        color = "black",
        size = .4,
        alpha = .8)  +
  geom_sf(data=stations_sf, fill="lightblue",shape=21, alpha=0.5)+
  theme_void()+
  coord_sf(xlim = c(9.9, 10.1),
          ylim = c(53.5, 53.65),
          expand = FALSE)+
  ggsave("test.png")

```

#identify most common trips
```{r}

#most common trips
common_trips <- trips %>% 
  left_join(stations, by=c("start_rental_zone_hal_id" = "station_id")) %>%
  left_join(stations, by=c("end_rental_zone_hal_id" = "station_id"),suffix=c("_start","_end")) %>%
  group_by(start_rental_zone_hal_id,end_rental_zone_hal_id) %>%
  mutate(number_trips = n()) %>%
  group_by(start_rental_zone_hal_id,end_rental_zone_hal_id,number_trips) %>%
  summarise() %>%
  rename(start_id = start_rental_zone_hal_id, end_id = end_rental_zone_hal_id) %>%
  arrange(desc(number_trips))

#show most common trips with names
common_trips %>%
  left_join(stations, by=c("start_id" = "station_id" )) %>%
    left_join(stations, by=c("end_id" = "station_id"),suffix=c("_start","_end")) %>%
  select(start_id, name_start, end_id,name_end, number_trips)

```

#identify most used stations 
```{r}
most_used_stations <- common_trips %>%
  group_by(end_id) %>%
  summarise(sum_ends = sum(number_trips)) %>%
  arrange(desc(sum_ends)) %>%
  left_join(stations, by=c("end_id" = "station_id"))

```
#from which landuse to which ?
```{r}
#which stations in which landuse?
#d <- st_join(stations_sf, landuse_sf)
# stations_nearest <- st_join(stations_sf, landuse_sf, join = st_nearest_feature)

library(units)
r250 <- set_units(250, m)

#do this only for five most used stations
stations_circles <- most_used_stations %>%
  #take only the most used stations (for "end")
  slice(1:20) %>%
  mutate(order = 1:n()) %>%
  # change to Irish grid, which uses meters
  st_as_sf(coords = c("lon", "lat"), crs=4326) %>%
  st_transform(.,29902) %>%
  #draw circle of 250m around each station
  st_buffer(., r250) %>%
  st_transform(.,4326)

#both stations and stations with circles are now back in 4326
#landuse_sf is also in 4326

#check points
ggplot() +
  geom_sf(data=landuse_sf, aes(fill=fclass), lwd = 0) +
  geom_sf(data = stations_sf)+
  geom_sf(data=stations_circles,alpha=0.1)+
  coord_sf(xlim = c(9.95, 10.05),
          ylim = c(53.54, 53.56),
          expand = FALSE)


#now which landuse polygons intersect with our station radius?
stations_circles_nearest <- st_intersection(landuse_sf, stations_circles)

#get a percentage of surrounding amendities (=landuse categories) for each station)
stations_nearest_amenities <- stations_circles_nearest %>%
  as.data.frame() %>%
  group_by(end_id,fclass,name.1,order) %>%
  summarise(n_type=n()) %>%
  group_by(end_id) %>%
  mutate(perc=n_type/sum(n_type)*100) %>%
  arrange(order)

ggplot(stations_nearest_amenities,aes(y=n_type, x=factor(name.1,levels=unique(stations_nearest_amenities$name.1)),fill=fclass)) +
  geom_bar(position="dodge", stat="identity")+
  theme_classic()+
  ggtitle("Counts of nearby amenties (based on counts of landuse data within 250m radius)\nShow only top 20 used stations")+
  labs(x="name",y="counts")+
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+
  ggsave("counts_ameneties_at_end_points.png",width=8,height = 6)
  


#now how many pubs intersect with our station radius?
query_restaurant <- opq(bbox = 'Hamburg') %>%
  add_osm_feature(key = 'restaurant') %>% 
  osmdata_sf()

query_restaurant
stations_circles
ggplot() +
  geom_sf(data = query_bar$osm_points,
          inherit.aes = FALSE,
          size = 1,
          alpha = 0.5,
          shape=21,
          fill="red")

stations_circles_nearest_restaurants <- st_intersection(query_restaurant, stations_circles)




```


```{r}
common_trips_landuse <- common_trips %>% 
  left_join(stations_nearest, by=c("start_id" = "station_id")) %>%
  left_join(stations_nearest, by=c("end_id" = "station_id"),suffix=c("_start","_end"))

counts_trips_landuse <- common_trips_landuse %>%
  group_by(fclass_start, fclass_end)%>%
  summarise(n=sum(number_trips)) %>%
  arrange(desc(n)) 

ggplot(counts_trips_landuse,aes(fclass_start,fclass_end))+
  geom_tile(aes(fill=n))+
  scale_fill_viridis(option="magma")

```


#make trips as sf line objects 
```{r}

make_line <- function(lat_start, lat_end, lon_start, lon_end) {
      st_linestring(matrix(c(lat_start, lat_end, lon_start, lon_end), 2, 2))
}


five_common_trips <- common_trips[1:5,] %>%
  left_join(stations, by=c("start_id" = "station_id")) %>%
  left_join(stations, by=c("end_id" = "station_id"),suffix=c("_start","_end")) %>% 
  ungroup() %>%
  # select(lat_start, lon_start, lat_end, lon_end) %>%
  mutate(line=map(make_line,lat_start, lat_end, lon_start, lon_end))
  # st_as_sfc(crs = 4326)

ggplot() +
  geom_sf(data = streets$osm_lines,
        inherit.aes = FALSE,
        color = "black",
        size = .4,
        alpha = .8)  +
  geom_sf(data=common_trips_sf_e, alpha=0.75,aes(color=count_ends))+
  theme_void()+
  coord_sf(xlim = c(9.75, 10.2),
          ylim = c(53.5, 53.65),
          expand = FALSE)+
  scale_color_viridis(option="magma")+
  ggsave("most_common_end_points.png",width=8,height = 4)
```

```{r}
common_trips_sf_s <- common_trips %>%
  group_by(start_id) %>%
  mutate(count_starts = sum(number_trips)) %>%
  group_by(start_id,count_starts) %>%
  summarise() %>%
  left_join(stations, by=c("start_id" = "station_id")) %>% 
  arrange(desc(count_starts)) %>%
  st_as_sf(coords = c("lon", "lat"), crs=4326)

common_trips_sf_e <- common_trips %>%
  group_by(end_id) %>%
  mutate(count_ends = sum(number_trips)) %>%
  group_by(end_id,count_ends) %>%
  summarise() %>%
  left_join(stations, by=c("end_id" = "station_id")) %>%
  arrange(desc(count_ends)) %>%
  st_as_sf(coords = c("lon", "lat"), crs=4326)

ggplot() +
  geom_sf(data = streets$osm_lines,
        inherit.aes = FALSE,
        color = "black",
        size = .4,
        alpha = .8)  +
  geom_sf(data=common_trips_sf_s, alpha=0.75,aes(color=count_starts))+
  theme_void()+
  coord_sf(xlim = c(9.75, 10.2),
          ylim = c(53.5, 53.65),
          expand = FALSE)+
  scale_color_viridis(option="magma")+
  ggsave("most_common_start_points.png",width=8,height = 4)

ggplot() +
  geom_sf(data = streets$osm_lines,
        inherit.aes = FALSE,
        color = "black",
        size = .4,
        alpha = .8)  +
  geom_sf(data=common_trips_sf_e, alpha=0.75,aes(color=count_ends))+
  theme_void()+
  coord_sf(xlim = c(9.75, 10.2),
          ylim = c(53.5, 53.65),
          expand = FALSE)+
  scale_color_viridis(option="magma")+
  ggsave("most_common_end_points.png",width=8,height = 4)

```