from __future__ import division
import seaborn as sns
import sys
sys.path.append('/home/fabee/code/')
from matplotlib.pyplot import *
from fabee.Plotting import *
from scipy import stats
from numpy import *

sns.set_context("talk", font_scale=1.5, rc={"lines.linewidth": 2.5})

# ---------------------------------------------------------------------------
fig, ax = subplots()
fig.subplots_adjust(bottom=.3, left=.3)
n = 50
x = loadtxt('scripts/thymusglandweights.dat')[:n]
ax.bar([0,1],[mean(x),mean(x)],yerr = [std(x,ddof=1), std(x,ddof=1)/sqrt(n)],
       facecolor='dodgerblue', alpha=.8,width=.7, align='center',
       error_kw={'color':'k','lw':2}, capsize=10, ecolor='k')
ax.set_title('standard deviation or standard error?',fontsize=14, fontweight='bold')

ax.set_xlim([-.5,1.5])
box_off(ax)
#disjoint_axes(ax)
ax.set_xticks([0,1])
ax.set_xticklabels([r'$\hat\sigma$', r'$\frac{\hat\sigma}{\sqrt{n}}$'], fontsize=30)

ax.set_ylabel(r'$\frac{1}{n}\sum_{i=1}^n x_i$',fontsize=30, fontweight='bold')

fig.savefig('figs/StandardErrorOrStandardDeviation.pdf')

# ---------------------------------------------------------------------------
fig, ax = subplots()

t = linspace(-5,5,1000)
t2 = linspace(stats.laplace.ppf(0.025),stats.laplace.ppf(1-0.025),1000)

ax.fill_between(t,stats.laplace.pdf(t),color='dodgerblue')
ax.set_xticks([])
ax.text(5,-0.05, r'$\hat m$',fontsize=30)
ax.text(0,0.7, r'$m$',fontsize=30)
ax.set_yticks([])
#disjoint_axes(ax)
box_off(ax)

ax.set_title('putative sampling distribution of the median',fontsize=14, fontweight='bold')
ax.axis([-5,5,0,.8])
ax.plot([0,0],[0,.7],'--k',lw=2)

fig.savefig('figs/samplingDistributionMedian00.pdf')

ax.fill_between(t2,stats.laplace.pdf(t2),color='crimson')

fig.savefig('figs/samplingDistributionMedian01.pdf')

# ---------------------------------------------------------------------------
fig, ax = subplots()
k = 7
N = 21
F = stats.f
t = linspace(1e-6,8,1000)
t2= linspace(F.ppf(0.95,k-1,N-k),8,1000)

ax.fill_between(t,F.pdf(t,k-1,N-k),color='dodgerblue')
ax.fill_between(t2,F.pdf(t2,k-1,N-k),color='crimson')
ax.set_xlabel('group MS/ error MS')
ax.set_ylabel(r'p(group MS/ error MS| $H_0$)')
ax.set_title('F-distribution',fontsize=14, fontweight='bold')
ax.set_ylim((0,0.8))
box_off(ax)
fig.savefig('figs/Fdistribution00.pdf')

# ---------------------------------------------------------------------------
fig, ax = subplots()
n = 5
p = stats.t.pdf
t = linspace(-5,8,1000)
t0 = 1.5
t00 = 1.

mu0 = 3
t1 = linspace(-5,t00,1000)
t2 = linspace(t0,8,1000)
t3 = linspace(-5,-t0,1000)
ax.fill_between(t,p(t,n-1),color='dodgerblue',alpha=1)
ax.fill_between(t2,p(t2,n-1),color='indigo',alpha=1)
ax.fill_between(t3,p(t3,n-1),color='indigo',alpha=1)
ax.set_xlabel('t')
ax.set_ylabel(r'sampling distribution')
ax.set_ylim((0,0.8))
box_off(ax)
fig.savefig('figs/experimentalDesign00.pdf')


ax.fill_between(t,p(t,n-1,loc=mu0),color='lime',alpha=.5)
ax.fill_between(t1,p(t1,n-1,loc=mu0),color='magenta',alpha=1)
ax.arrow(0,.4,mu0,0,head_width=0.05)
ax.arrow(mu0,.4,-mu0,0,head_width=0.05)
ax.text(mu0/2,.45,r'$\delta$',fontsize=20)
ax.set_xlabel('t')
ax.set_ylabel(r'sampling distribution')
ax.set_ylim((0,0.8))
box_off(ax)
fig.savefig('figs/experimentalDesign01.pdf')