In [4]:

    

from py2neo import Graph, Node, Relationship


    

In [62]:

    

## Connecting to graph database and creating a graph of cities
## I am using default user `neo4j` and password is changed to `admin`
g = Graph("http://neo4j:admin@localhost:7474")


    

In [63]:

    

type(g)


    

    
Out[63]:





py2neo.database.Graph

In [11]:

    

## Starting a transaction on graph and adding nodes to it
tx = g.begin()
with open('node_city.txt') as f_in:
    for line in f_in:
        city_list = [x.strip() for x in line.rstrip().split(',')]
        city = Node("City", name=city_list[0], population=int(city_list[1]))
        tx.create(city)
tx.commit()


    

In [76]:

    

## Running a command on Graph
for node in g.run("MATCH (a:City) RETURN a.name, a.population"):
    print(node)


    

    




<Record a.name='New_York' a.population=8491079>
<Record a.name='Los_Angeles' a.population=3928864>
<Record a.name='Chicago' a.population=2722389>
<Record a.name='Houston' a.population=2239558>
<Record a.name='Philadelphia' a.population=1560297>
<Record a.name='Phoenix' a.population=1537058>
<Record a.name='San_Antonio' a.population=1436697>
<Record a.name='San_Diego' a.population=1381069>
<Record a.name='Dallas' a.population=1281047>
<Record a.name='San_Jose' a.population=1015785>
<Record a.name='Austin' a.population=912791>
<Record a.name='Jacksonville' a.population=853382>
<Record a.name='San_Francisco' a.population=852469>
<Record a.name='Indianapolis' a.population=848788>
<Record a.name='Columbus' a.population=835957>
<Record a.name='Fort_Worth' a.population=812238>
<Record a.name='Charlotte' a.population=809958>
<Record a.name='Detroit' a.population=680250>
<Record a.name='Seattle' a.population=66342>
<Record a.name='Denver' a.population=663862>

In [95]:

    

##Select nodes of this graph
nodes = g.nodes


    

In [99]:

    

## Making relationship between nodes
tx = g.begin()
with open('edge_distance.txt') as f_in:
    for line in f_in:
        edge_list = [x.strip() for x in line.split(',')]
        city1_name = edge_list[0]
        city2_name = edge_list[1]
        city1_node = nodes.match(name=city1_name)
        city2_node = nodes.match(name=city2_name)
        city_pair = Relationship(city1_node.first(), 'FLIGHT_BETWEEN', city2_node.first())
        city_pair['distance'] = edge_list[2]
        tx.create(city_pair)
tx.commit()


    

In [104]:

    

##importing pandas for analysis
import pandas as pd


    

In [112]:

    

## Importing relationship matcher
from py2neo import RelationshipMatcher


    

In [113]:

    

## Creating RelationMatcher on graph
rm = RelationshipMatcher(g)


    

In [115]:

    

## Collecting relationships in list
flights = list()
for cities in rm.match(r_type='FLIGHT_BETWEEN'):
    city1_name = cities.nodes[0]['name']
    city2_name = cities.nodes[1]['name']
    distance = cities['distance']
    flights.append([city1_name, city2_name, distance])


    

In [117]:

    

## Creating Pandas DataFrame
flights_df = pd.DataFrame(flights)


    

In [118]:

    

## Viewing first 5 elements in dataframe
flights_df.head()


    

    
Out[118]:







  

    

      

      
0
      
1
      
2
    
  
  

    

      
0
      
New_York
      
San_Francisco
      
2572
    
    

      
1
      
New_York
      
Dallas
      
1373
    
    

      
2
      
New_York
      
Philadelphia
      
95
    
    

      
3
      
New_York
      
Chicago
      
733
    
    

      
4
      
New_York
      
Los_Angeles
      
2451