anup.py

#Importing data testing network graph for bigger number of records
import networkx as nx
import pandas as pd
import numpy as np
df =pd.read_sas("C:\\My Codes\\SNA_New\\testdata3.sas7bdat",format='SAS7BDAT',encoding="ISO-8859-15")
#print(df)
df1 =pd.read_sas("C:\\My Codes\\SNA_New\\min_nbr_v3.sas7bdat",format='SAS7BDAT',encoding="ISO-8859-15")

L=df1.values.tolist()
#L
G5=nx.from_pandas_edgelist(df,'calling_num','called_num',create_using=nx.DiGraph())
nx.is_directed(G5)
G6=G5.to_undirected()
e=nx.non_edges(G6)
flat_set=set()
for sublist in L:
    for item in sublist:
        flat_set.insert(item)
		
L1=nx.jaccard_coefficient(G6)
L2=((u,v,z) for (u,v,z) in L1 if z>0)
L4=((u,v,z) for (u,v,z) in L2 if u in flat_set or v in flat_set)

import time
start=time.time()

import csv
with open("C:\\My Codes\\SNA_New\\jac_coeff_020419.csv","w+",newline="\n") as f:
    n = 0
    s = ""
    start1=time.time()
    for rec in L4:
        n = n+1
        s = s + str(rec[0]) + ", " + str(rec[1]) + ", " + str(rec[2]) + "\n" #one can optimize this further.
        
        if (n == 10): #Max records to write at a time
            end1=time.time()
            print("string conversion",end1-start1) #string
            start2=time.time()
            f.write(s)
            end2=time.time()
            print("f.write",end2-start2) #string
            start3=time.time()
            f.flush()
            end3=time.time()
            print("f.flush",end3-start3) #string
            n = 0
            s = ""
            break
    f.write(s)
    
end=time.time()
print("overall",end-start)