Exploratory Data Analysis - Solution¶

Make sure the libraries are imported by executing this code (remember to unset Reset all runtimes before running):

In [2]:

from py2neo import Graph
import pandas as pd

import matplotlib 
import matplotlib.pyplot as plt

plt.style.use('fivethirtyeight')
pd.set_option('display.float_format', lambda x: '%.3f' % x)

Update the cell below to use the IP Address, Bolt Port, and Password, as you did previously.

In [6]:

# Change the line of code below to use the Bolt URL,  and Password of your Neo4j Database instance.
# graph = Graph("<Bolt URL>", auth=("neo4j", "<Password>")) 
 
graph = Graph("bolt://localhost:7687", auth=("neo4j", "letmein"))

Exercise¶

Create a similar query showing the articles that cite the most other papers? Note that you will only need to do a slight tweak of the Cypher query to do this.

In [4]:

query = """
MATCH (a:Article)
RETURN size((a)-[:CITED]->()) AS cited
"""

cited_df = graph.run(query).to_data_frame()
cited_df.describe([.25, .5, .75, .9, .99])

Out[4]:

	cited
count	51956.000
mean	0.553
std	1.301
min	0.000
25%	0.000
50%	0.000
75%	1.000
90%	2.000
99%	6.000
max	51.000

In [ ]:

fig1, ax1 = plt.subplots()
ax1.hist(pd.Series(cited_df['cited'].dropna()), 1250, density=True, facecolor='g', alpha=0.75)
plt.tight_layout()
plt.show()

What about a distribution of the number of articles published by authors?

In [5]:

query = """
MATCH (a:Author)
RETURN size((a)<-[:AUTHOR]-()) AS published
"""

published_df = graph.run(query).to_data_frame()
published_df.describe([.25, .5, .75, .9, .99])

Out[5]:

	published
count	80299.000
mean	1.751
std	2.064
min	1.000
25%	1.000
50%	1.000
75%	2.000
90%	3.000
99%	10.000
max	89.000