Révision 13c68160
| b/papers/2014/reservation/experiments/src/python/lib/kadeploy.py | ||
|---|---|---|
| 7 | 7 |
import numpy as np |
| 8 | 8 |
from matplotlib import rc |
| 9 | 9 |
from pylab import * |
| 10 |
from scipy.stats import gaussian_kde, lognorm, rayleigh, norm |
|
| 10 |
from scipy.stats import gaussian_kde, lognorm, rayleigh, norm, kstest, ks_2samp
|
|
| 11 | 11 |
import scipy.stats as stats |
| 12 | 12 |
rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})
|
| 13 | 13 |
rc('text', usetex=True)
|
| 14 | 14 |
plt.rcParams.update({'font.size': 14})
|
| 15 | 15 |
from datetime import datetime |
| 16 |
from statsmodels.stats.gof import powerdiscrepancy |
|
| 16 | 17 |
|
| 17 | 18 |
TZINFO = pytz.timezone("Europe/Paris")
|
| 18 | 19 |
|
| ... | ... | |
| 157 | 158 |
ar1 = np.array(values) |
| 158 | 159 |
fig, ax = plt.subplots() |
| 159 | 160 |
|
| 160 |
num_bins = 50
|
|
| 161 |
num_bins = 60
|
|
| 161 | 162 |
n, bins, patches = plt.hist(ar1, bins=num_bins, histtype='bar', |
| 162 | 163 |
normed=True, facecolor='#708090', alpha=0.5, rwidth=0.8) |
| 163 | 164 |
plt.subplots_adjust(left=0.15) |
| 164 | 165 |
|
| 165 | 166 |
max = round(bins[len(bins) - 1]) |
| 166 |
density = gaussian_kde(ar1) |
|
| 167 |
xs = np.linspace(0, max, 300) |
|
| 168 |
density.covariance_factor = lambda : .25 |
|
| 169 |
density._compute_covariance() |
|
| 167 |
# xs = np.linspace(0, max, len(values)) |
|
| 168 |
xs = np.linspace(0, max, len(values)) |
|
| 169 |
|
|
| 170 |
# density = gaussian_kde(ar1) |
|
| 171 |
# density.covariance_factor = lambda : .25 |
|
| 172 |
# density._compute_covariance() |
|
| 170 | 173 |
# plt.plot(xs, density(xs), 'r--') |
| 171 | 174 |
|
| 172 | 175 |
ax.set_xlabel(x_label) |
| ... | ... | |
| 190 | 193 |
|
| 191 | 194 |
# dist_names = ['lognorm', 'beta', 'betaprime', 'gamma', 'rayleigh'] |
| 192 | 195 |
dist_names = ['lognorm', 'beta', 'betaprime', 'gamma', 'rayleigh'] |
| 196 |
#dist_names = ['lognorm', 'beta', 'betaprime', 'gamma'] |
|
| 197 |
# dist_names = ['beta', 'betaprime', 'gamma'] |
|
| 198 |
ks_tests = {}
|
|
| 193 | 199 |
|
| 194 | 200 |
for dist_name in dist_names: |
| 195 | 201 |
dist = getattr(stats, dist_name) |
| 196 | 202 |
param = dist.fit(ar1) |
| 197 | 203 |
pdf_fitted = dist.pdf(xs, *param[:-2], loc=param[-2], scale=param[-1]) |
| 204 |
|
|
| 205 |
kst = kstest(ar1, dist_name, param) |
|
| 206 |
ks_tests[dist_name] = kst |
|
| 207 |
|
|
| 198 | 208 |
plt.plot(xs, pdf_fitted, label=dist_name) |
| 199 | 209 |
plt.xlim(0, max) |
| 200 |
plt.legend(loc='upper right') |
|
| 201 | 210 |
|
| 202 |
# param = lognorm.fit(ar1) |
|
| 203 |
# pdf_fitted = lognorm.pdf(xs, param[0], loc=param[1], scale=param[2]) |
|
| 204 |
# plt.plot(xs, pdf_fitted, 'b') |
|
| 205 |
|
|
| 206 |
# param = norm.fit(ar1) |
|
| 207 |
# pdf_fitted = norm.pdf(xs, loc=param[0], scale=param[1]) |
|
| 208 |
# plt.plot(xs, pdf_fitted, 'b') |
|
| 211 |
plt.legend(loc='upper right') |
|
| 209 | 212 |
|
| 210 | 213 |
#print param |
| 214 |
for key, value in ks_tests.iteritems(): |
|
| 215 |
print key, value |
|
| 211 | 216 |
|
| 212 |
# plt.show()
|
|
| 213 |
plt.savefig(out_file) |
|
| 217 |
plt.show() |
|
| 218 |
# plt.savefig(out_file)
|
|
| 214 | 219 |
|
| 215 | 220 |
|
| 216 | 221 |
def plot_graphs(): |
Formats disponibles : Unified diff