Modeling - Trial 01.Rmd

---
title: "Modeling - Trial 01"
author: "Denis Molin"
date: "28 novembre 2016"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## Query simple

Pour commencer, une petite requête qui récupère et affiche:
- les données d'une journée de la semaine: au hasard, le lundi.
- des stations situées à moins de 100m de l'adresse suivante: "8 boulevard saint michel, paris" (adresse purement arbitraire)

```{r cars}
# on definit le nombre de stations
max_distance   <- 200
# on definit l'adresse
address        <- "8 boulevard saint michel, paris"
# on definit le jour
jour           <- "Mon"
# on definit nb stations max
nb_stations_max <- 6
# on recupere les identifiants des 5
require(ggmap)
my.coordinates <- as.data.frame(geocode(address))
x<-paste0('{ "loc" : { "$near" : { "$geometry" :  { "type" : "Point" , "coordinates" : [ ',my.coordinates$lon, ' , ', my.coordinates$lat,'] } ,  "$maxDistance" : ',max_distance,'}}}')
require(mongolite)
m      <-mongo(collection = "velib",  url = paste0("mongodb://",my.env$user_mongo,":",my.env$pwd_mongo,"@thinkr.fr/velib"),verbose = TRUE, db = "velib")

# geoNear permet de requêter suivant l'index 2dsphere avec aggregation (https://docs.mongodb.com/v3.2/reference/operator/aggregation/geoNear/)
my.aggregate <- paste0('[
   {"$geoNear" : {
        "near": { "type": "Point", "coordinates": [',my.coordinates$lon, ' , ', my.coordinates$lat,'] },
        "distanceField": "dist.calculated",
        "maxDistance": ',max_distance,',
        "includeLocs": "dist.location",
        "num" : ',nb_stations_max,',
        "spherical": true
     }
   },
    { "$unwind": "$serie"},
    { "$match" : {"serie.day" : "',jour,'"}},
    { "$project" : {"_id": 0, "serie.hour" : 1, "serie.available_bikes":1 , "serie.available_bike_stands":1,"number": 1}} 
]') 

output <- m$aggregate(my.aggregate)
output$number <- factor(output$number)
require(ggplot2)
ggplot(output)+aes(x=serie$hour,y=serie$available_bike_stands,col=number)+geom_point()+facet_wrap(~number,nrow=2)+xlab('hour')+ylab('# available bikes')+ggtitle(label=paste0(jour, " : ",address))

```