In [ ]:

    

import networkx as nx


    

In [ ]:

    

import matplotlib
import matplotlib.pyplot as plt


    

In [ ]:

    

from collections import defaultdict


    

In [ ]:

    

import numpy as np


    

In [ ]:

    

import scipy.stats


    

In [ ]:

    

SOCIAL_GRAPH_FILENAME = 'data/london/ground-truth.txt'
CASCADE_FILENAME = 'data/london/cascades-london-renum.txt'
INFOPATH_FORMAT = True
TITLE = 'London [REAL]'
fig_name = 'london-real'


    

In [ ]:

    

added_nodes = set()
G = nx.DiGraph()


    

In [ ]:

    

with open(SOCIAL_GRAPH_FILENAME, 'r') as f:
        for line in f:
            try:
                u1, u2, alpha = line[:-1].split(',')
                u1, u2 = map(int, [u1, u2])
                alpha = float(alpha)
                for u in [u1, u2]:
                    if u not in added_nodes:
                        added_nodes.add(u)
                        G.add_node(u)
                G.add_edge(u1, u2, weight=alpha)
            except ValueError:
                print line


    

In [ ]:

    

# nx.draw(G)
# plt.show()


    

In [ ]:

    

# Maps (node_i, node_j) -> [delay1, delay2, ...]
delay_dict = defaultdict(list)
delay_dict_false = defaultdict(list)


    

In [ ]:

    

with open(CASCADE_FILENAME, 'r') as f:
    for line in f:
        if line == '\n': break
            
    for line in f:
        cascade_str = []
        if INFOPATH_FORMAT:
            cascade_str = line.split(';')[-1]
        else:
            cascade_str = line
        cascade_lst = map(int, cascade_str.split(','))
        cascade = [(cascade_lst[i], cascade_lst[i+1]) for i in range(0, len(cascade_lst)-1, 2)]
        cascade.sort(key=lambda x: x[1]) # Sort by the time of the checkin
        for i in range(1, len(cascade)-1):
            node_i = cascade[i][0]
            neigh_of_i = G.neighbors(node_i)
            for j in range(i, len(cascade)):
                # If (i, j) is an edge in the graph, append delay
                node_j = cascade[j][0]
                if node_j in neigh_of_i:
                    delay_dict[(node_i, node_j)].append(cascade[j][1] - cascade[i][1])
                else:
                    delay_dict_false[(node_i, node_j)].append(cascade[j][1] - cascade[i][1])


    

In [ ]:

    

interarrival_ordered_keys_list = sorted(delay_dict.keys(), key=lambda x: len(delay_dict[x]), reverse=True)
#interarrival_false_ordered_keys_list = sorted(delay_dict_false.keys(), key=lambda x: len(delay_dict[x]), reverse=True)


    

In [ ]:

    

print len(delay_dict_false.keys())
print len(delay_dict.keys())


    

In [ ]:

    

trans_len_list = [len(x) for x in delay_dict.values()]
trans_len_list_false = [len(x) for x in delay_dict_false.values()]
fig = plt.figure()
plt.gca().set_position((.1, .3, .8, .6))
plt.hist(trans_len_list)
plt.grid()

plt.legend()
# plt.ylabel("PDF")
plt.xlabel("Number of transmissions over edges")
plt.suptitle("Histogram for number of transmissions over true edges\n%s" % TITLE)
trans_text = '''
Mean number of transmissions over true edges: %.4f
Median number of transmissions over true edges: %.4f
Mean number of transmissions over false edges: %.4f
Median number of transmissions over false edges: %.4f
Number of true edges = %d
Number of false edges = %d
''' % (np.mean(trans_len_list), 
       np.median(trans_len_list), 
       np.mean(trans_len_list_false), 
       np.median(trans_len_list_false),
       len(delay_dict.keys()),
       len(delay_dict_false.keys()))
fig.text(0.1, 0.000001, trans_text)

plt.savefig('/home/tribhu/Dropbox/Development/ETH/pml/graphs/%s-transmissions.png' % (fig_name))
plt.show()

#print 'Mean number of transmissions over true edges: ', np.mean(trans_len_list)
#print 'Median number of transmissions over true edges: ', np.median(trans_len_list)
#print 'Mean number of transmissions over false edges: ', np.mean(trans_len_list_false)
#print 'Median number of transmissions over false edges: ', np.median(trans_len_list_false)


    

In [ ]:

    

# xbins = range(0,400)
x = np.linspace(0, 300, 100)


    

In [ ]:

    

n1, n2 = interarrival_ordered_keys_list[5]
lambda_ = G[n1][n2]['weight']
print lambda_


    

In [ ]:

    

colors = matplotlib.rcParams['axes.color_cycle']
expo = scipy.stats.expon
plt.plot(x, expo.pdf(x, scale=lambda_), lw=2, color=colors[2], label = "$\\alpha_{i,j} = %.3f$"%lambda_)
plt.fill_between(x, expo.pdf(x, scale=lambda_), color=colors[2], alpha = .33)

fig = plt.hist(delay_dict[(n1, n2)], bins=range(0,400), normed=True)

plt.grid()

plt.legend()
plt.ylabel("PDF")
plt.xlabel("Inter-infection time (in Days)")
plt.suptitle("Inter-infection time across edges for a fixed transmission rate\n%s" % TITLE);


    

In [ ]:

    

# plt.show()


    

In [ ]:

    

plt.savefig('/home/tribhu/Dropbox/Development/ETH/pml/graphs/%s.png' % (fig_name))


    

In [ ]:

    

plt.show()


    

In [ ]: