"""
use pandas and cubes olap to do the DSIM data statitics
Author Jinjun Sun
"""
from dsimdata import csv2data,openfilegui
import pandas as ps
import numpy as np
if __name__ == '__main__':
#csvfile = openfilegui()
csvfile = 'datasrc/2011_netjjs2sql.csv'
datadic = csv2data(csvfile)
dataseries = ps.Series(datadic)
print np.log(map(float, dataseries[2][10:20]))
typelist = []
inumber = len(datadic['incident_duration'])
typenamestr = 'R_incident_manager_id'
for i in range(inumber):
typename = datadic[typenamestr][i] #+
#typename = datadic['R_incident_subtype'][i]
typelist.append(typename)
typelist = list(set(typelist))
for i in typelist:
Y[i] = []
for i in range(inumber):
typename = datadic[typenamestr][i]#+
#typename = datadic['R_incident_subtype'][i]
incdur = float(datadic['incident_duration'][i])/60
if incdur <= 5:
Y[typename].append(incdur)
for j in range(len(typelist)):
figure(j)
n, bins, patches = hist(Y[typelist[j]], 50, normed=1, facecolor = 'blue')
title("key for the hist %s"% typelist[j])
show()
return 0
if __name__ == "__main__":
filename = 'datasrc/bravo6to18.csv'
datadic = csv2data(filename)
n = bravotype(datadic)
