source: trunk/bin/graph @ 10

#!/usr/bin/env python3

import sys
import os
import time
import random
import tempfile
import argparse
import socket
import json

import numpy
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt


VERSION = "{DEVELOPMENT}"
if VERSION == "{DEVELOPMENT}":
    script_dir = '.'
    try:
        script_dir = os.path.dirname(os.path.realpath(__file__))
    except:
        try:
            script_dir = os.path.dirname(os.path.abspath(sys.argv[0]))
        except:
            pass
    sys.path.append("%s/../lib" % script_dir)

from nanownlib import *
from nanownlib.stats import *
import nanownlib.storage


parser = argparse.ArgumentParser(
    description="")
parser.add_argument('db_file', default=None,
                    help='')
options = parser.parse_args()
db = nanownlib.storage.db(options.db_file)


def differences(db, unusual_case, column='packet_rtt'):
    cursor = db.conn.cursor()
    query="""
      SELECT %(column)s-(SELECT avg(%(column)s) FROM probes,analysis
                         WHERE analysis.probe_id=probes.id AND probes.test_case!=:unusual_case AND probes.type in ('train','test') AND sample=u.sample)
      FROM (SELECT probes.sample,%(column)s FROM probes,analysis 
                         WHERE analysis.probe_id=probes.id AND probes.test_case =:unusual_case AND probes.type in ('train','test')) u
      """ % {"column":column}
    params = {"unusual_case":unusual_case}
    cursor.execute(query, params)
    for row in cursor:
        yield row[0]


def timeSeries(db, probe_type, unusual_case):
    cursor = db.conn.cursor()
    query="""
      SELECT time_of_day,packet_rtt AS uc,(SELECT avg(packet_rtt) FROM probes,analysis
                                           WHERE analysis.probe_id=probes.id AND probes.test_case!=:unusual_case AND probes.type=:probe_type AND sample=u.sample) AS oc
      FROM (SELECT time_of_day,probes.sample,packet_rtt FROM probes,analysis 
                                           WHERE analysis.probe_id=probes.id AND probes.test_case =:unusual_case AND probes.type=:probe_type) u
    """
    
    params = {"probe_type":probe_type,"unusual_case":unusual_case}
    cursor.execute(query, params)
    for row in cursor:
        yield {'time_of_day':row['time_of_day'],unusual_case:row['uc'],'other_cases':row['oc']}
#samples,derived,null_derived = parse_data(input1)

#trust = trustValues(derived, sum)
#weights = linearWeights(derived, trust, 0.25)
#print('(test): %f' % weightedMean(derived,weights))

diffs = list(differences(db, 'long'))
reported_diffs = list(differences(db, 'long', 'reported'))
#shorts = [s['packet_rtt'] for s in samples.values() if s['test_case']=='short']
#longs = [s['packet_rtt'] for s in samples.values() if s['test_case']=='long']

short_overtime = [(sample['time_of_day'],sample['short']) for sample in timeSeries(db,'train','short')]
long_overtime = [(sample['time_of_day'],sample['long']) for sample in timeSeries(db,'train','long')]
diff_overtime = [(sample['time_of_day'],sample['long']-sample['other_cases']) for sample in timeSeries(db,'train','long')]
short_overtime.sort()
long_overtime.sort()
diff_overtime.sort()

print('packet_rtt diff median: %f' % statistics.median(diffs))
print('packet_rtt diff midhinge: %f' % midsummary(diffs))
print('packet_rtt diff trimean: %f' % trimean(diffs))
print('packet_rtt diff quadsummary: %f' % quadsummary(diffs))
print('packet_rtt diff ubersummary: %f' % ubersummary(diffs))
print('packet_rtt diff MAD: %f' % mad(diffs))
print('reported diff trimean: %f' % trimean(reported_diffs))
print('reported diff quadsummary: %f' % quadsummary(reported_diffs))
print('reported diff ubersummary: %f' % ubersummary(reported_diffs))
print('reported diff MAD: %f' % mad(reported_diffs))

import cProfile
kresults = kfilter({},diffs)
#print('packet_rtt diff kfilter: ', numpy.mean(kresults['est']), kresults['var'])
print('packet_rtt diff kfilter: ', kresults['est'][-1], kresults['var'][-1])
kresults = kfilter({},reported_diffs)
#print('reported diff kfilter: ', numpy.mean(kresults['est']), kresults['var'][-1])
print('reported diff kfilter: ', kresults['est'][-1], kresults['var'][-1])


#all_data = longs+shorts
#all_data.sort()
#cut_off_low = all_data[0]
#cut_off_high = all_data[int(len(all_data)*0.997)]


plt.clf()
plt.title("Packet RTT over time")
plt.xlabel('Time of Day')
plt.ylabel('RTT')
s = plt.scatter([t for t,rtt in short_overtime], [rtt for t,rtt in short_overtime], s=1, color='red', alpha=0.6)
l = plt.scatter([t for t,rtt in long_overtime], [rtt for t,rtt in long_overtime], s=1, color='blue', alpha=0.6)
d = plt.scatter([t for t,rtt in diff_overtime], [rtt for t,rtt in diff_overtime], s=1, color='purple', alpha=0.6)
plt.legend((s,l,d), ('short','long','difference'), scatterpoints=1)
#plt.savefig('paper/figures/comcast-powerboost1.png')
plt.show()

short_overtime,long_overtime,diff_overtime = None,None,None


num_bins = 300
reported_diffs.sort()
cut_off_low = reported_diffs[int(len(diffs)*0.003)]
cut_off_high = reported_diffs[int(len(diffs)*0.997)]

plt.clf()
# the histogram of the data
n, bins, patches = plt.hist(reported_diffs, num_bins, normed=1, color='black', histtype='step', alpha=0.8,
                            range=(cut_off_low,cut_off_high))
plt.xlabel('RTT Difference')
plt.ylabel('Probability')
plt.title(r'Histogram - distribution of differences')

# Tweak spacing to prevent clipping of ylabel
plt.subplots_adjust(left=0.15)
#plt.legend()
plt.show()
#plt.savefig('paper/graphs/dists-vs-dist-of-diffs2.svg')




num_bins = 300
diffs.sort()
cut_off_low = diffs[int(len(diffs)*0.003)]
cut_off_high = diffs[int(len(diffs)*0.997)]

plt.clf()
# the histogram of the data
n, bins, patches = plt.hist(diffs, num_bins, normed=1, color='purple', histtype='step', alpha=0.8,
                            range=(cut_off_low,cut_off_high))
plt.xlabel('RTT Difference')
plt.ylabel('Probability')
plt.title(r'Histogram - distribution of differences')

# Tweak spacing to prevent clipping of ylabel
plt.subplots_adjust(left=0.15)
#plt.legend()
plt.show()
#plt.savefig('paper/graphs/dists-vs-dist-of-diffs2.svg')

sys.exit(0)



num_bins = 150
# the histogram of the data
n, bins, patches = plt.hist((shorts,longs), num_bins, normed=1, label=['short', 'long'], color=['red','blue'], histtype='step', alpha=0.8,
                            range=(cut_off_low,cut_off_high))
#n, bins, patches = plt.hist(shorts2+longs2, num_bins, normed=1, facecolor='blue', histtype='step', alpha=0.3)
# add a 'best fit' line
#y = mlab.normpdf(bins, mu, sigma)
#plt.plot(bins, y, 'r--')
plt.xlabel('packet_rtt')
plt.ylabel('Probability')
plt.title(r'Histogram - RTT short and long')

# Tweak spacing to prevent clipping of ylabel
plt.subplots_adjust(left=0.15)
plt.legend()
#plt.show()
plt.savefig('paper/figures/comcast-powerboost2.svg')




num_trials = 200


subsample_sizes = (50,150,300,500,700,1000,2000,3000,5000,7000,10000,15000,20000)
estimator = functools.partial(boxTest, 0.07, 0.08)
performance = []
for subsample_size in subsample_sizes:
    estimates = bootstrap(derived, subsample_size, num_trials, estimator)
    performance.append(100.0*len([e for e in estimates if e == 1])/num_trials)

null_performance = []
for subsample_size in subsample_sizes:
    null_estimates = bootstrap(null_derived, subsample_size, num_trials, estimator)
    null_performance.append(100.0*len([e for e in null_estimates if e == 0])/num_trials)

plt.clf()
plt.title("boxTest bootstrap")
plt.xlabel('sample size')
plt.ylabel('performance')
plt.scatter(subsample_sizes, performance, s=2, color='red', alpha=0.6)
plt.scatter(subsample_sizes, null_performance, s=2, color='blue', alpha=0.6)
plt.show()



subsample_sizes = (50,150,300,400,500,700,1000,2000,3000,4000,5000,7000,10000)
estimator = diffMedian
performance = []
for subsample_size in subsample_sizes:
    estimates = bootstrap(derived, subsample_size, num_trials, estimator)
    performance.append(100.0*len([e for e in estimates if e > expected_mean*0.9 and e < expected_mean*1.1])/num_trials)

plt.clf()
plt.title("diff median bootstrap")
plt.xlabel('sample size')
plt.ylabel('performance')
plt.scatter(subsample_sizes, performance, s=1, color='red', alpha=0.6)
plt.show()




subsample_sizes = (50,150,300,400,500,700,1000,2000,3000,4000,5000,7000,10000)
weight_funcs = (linearWeights, prunedWeights)
for wf in weight_funcs:
    estimator = functools.partial(estimateMean, hypotenuse, wf, 0.40)
    performance = []
    for subsample_size in subsample_sizes:
        estimates = bootstrap(derived, subsample_size, num_trials, estimator)
        performance.append(100.0*len([e for e in estimates if e > expected_mean*0.9 and e < expected_mean*1.1])/num_trials)

    plt.clf()
    plt.title(repr(wf))
    plt.xlabel('sample size')
    plt.ylabel('performance')
    plt.scatter(subsample_sizes, performance, s=1, color='red', alpha=0.6)
    plt.show()



num_bins = 300
# the histogram of the data
n, bins, patches = plt.hist((tsshorts,tslongs), num_bins, normed=1, label=['short', 'long'], color=['red','blue'], histtype='step', alpha=0.8)
#n, bins, patches = plt.hist(shorts2+longs2, num_bins, normed=1, facecolor='blue', histtype='step', alpha=0.3)
# add a 'best fit' line
#y = mlab.normpdf(bins, mu, sigma)
#plt.plot(bins, y, 'r--')
plt.xlabel('packet_rtt')
plt.ylabel('Probability')
plt.title(r'Histogram - tsval_rtt short vs long')

# Tweak spacing to prevent clipping of ylabel
plt.subplots_adjust(left=0.15)
plt.legend()
plt.show()



    
####
#trust_methods = [min,max,sum,difference,product]
trust_methods = [sum,product,hypotenuse]
colors = ['red','blue','green','purple','orange','black']
weight_methods = [prunedWeights, linearWeights]
alphas = [i/100.0 for i in range(0,100,2)]




plt.clf()
plt.title(r'Trust Method Comparison - Linear')
plt.xlabel('Alpha')
plt.ylabel('Mean error')
paths = []
for tm in trust_methods:
    trust = trustValues(derived, tm)
    series = []
    for alpha in alphas:
        weights = linearWeights(derived, trust, alpha)
        series.append(weightedMean(derived, weights) - expected_mean)

    paths.append(plt.scatter(alphas, series, s=1, color=colors[len(paths)],alpha=0.6))

plt.legend(paths, [repr(tm) for tm in trust_methods], scatterpoints=1)
plt.show()



plt.clf()
plt.title(r'Trust Method Comparison - Pruned')
plt.xlabel('Alpha')
plt.ylabel('Mean error')
paths = []
for tm in trust_methods:
    trust = trustValues(derived, tm)
    series = []
    for alpha in alphas:
        weights = prunedWeights(derived, trust, alpha)
        series.append(weightedMean(derived, weights) - expected_mean)

    paths.append(plt.scatter(alphas, series, s=1, color=colors[len(paths)],alpha=0.6))

plt.legend(paths, [repr(tm) for tm in trust_methods], scatterpoints=1)
plt.show()


sys.exit(0)

plt.clf()
plt.title(r'Trust Method Comparison - Inverted')
plt.xlabel('Alpha')
plt.ylabel('Mean error')
paths = []
for tm in trust_methods:
    trust = trustValues(derived, tm)
    series = []
    for alpha in alphas:
        weights = invertedWeights(derived, trust, alpha)
        series.append(weightedMean(derived, weights) - expected_mean)

    paths.append(plt.scatter(alphas, series, s=1, color=colors[len(paths)],alpha=0.6))

plt.legend(paths, [repr(tm) for tm in trust_methods], scatterpoints=1)
plt.show()


plt.clf()
plt.title(r'Trust Method Comparison - Arctangent')
plt.xlabel('Alpha')
plt.ylabel('Mean error')
paths = []
for tm in trust_methods:
    trust = trustValues(derived, tm)
    series = []
    for alpha in alphas:
        weights = arctanWeights(derived, trust, alpha)
        series.append(weightedMean(derived, weights) - expected_mean)

    paths.append(plt.scatter(alphas, series, s=1, color=colors[len(paths)],alpha=0.6))

plt.legend(paths, [repr(tm) for tm in trust_methods], scatterpoints=1)
plt.show()