# -*- coding: utf-8 -*-

"""

@author: Luz

"""

import sys

import numpy as np

import seaborn as sns

import matplotlib.pyplot as plt #for plots

#from functionscours8 import LSLassoCV, PredictionError, ScenarioEquiCor, ridge_path

from sklearn.linear_model import LassoCV, RidgeCV

#from functionscours8 import Refitting

import pandas as pd

from sklearn import linear_model

from sklearn.linear_model import lasso_path

#import statsmodels.api as sm

import os

###################################################################################

# Plot initialization

import matplotlib.pyplot as plt #for plots

dirname="../srcimages/"

dirname="./"

imageformat='.jpg'

plt.close('all')

################------ Exercice 2 ------###############################

# Load data

filew = 'winequality-white.csv'

filer = 'winequality-red.csv'

#read table data

dataw = pd.read_csv(filew,sep=';')

datar = pd.read_csv(filer,sep=';')

################------ Exercice 2.1 ------###############################

data = datar

noms = data.columns.values

#data.shape

y = data[noms[11]]

X = data[noms[0:11]]

(nsamples,nfeatures)=X.shape

################------ Exercice 2.1.a ------###############################

mymodel = linear_model.LinearRegression(fit_intercept=True)

mymodel.fit(X,y)

print("2.1.a ) Coefs modele lineaire + intercept")

print([mymodel.coef_ , mymodel.intercept_])

print "2.1.a) Nombre d'observations: ", y.shape[0]

################------ Exercice 2.1.b ------###############################

y_cr = (y-y.mean())/y.std()

X_cr = (X-np.mean(X))/np.std(X)

print "2.1.b) lignes 158, 159, 213 de X_cr\n", X_cr.iloc[[158,159,213]]

print "2.1.b) lignes 158, 159, 213 de y_cr\n", y_cr[[158,159,213]]

################------ Exercice 2.3 ------###############################

data = dataw

noms = data.columns.values

#data.shape

y = data[noms[11]]

X = data[noms[0:11]]

(nsamples,nfeatures)=X.shape

y_cr = (y-y.mean())/y.std()

X_cr = (X-np.mean(X))/np.std(X)

def simpleLasso(X,y,lambdapar,fit_inter):

return clf.coef_, Yhat

npoints = 25

lend = 1

eps = 1e-4

lstart = lend*eps

alphas = np.logspace(np.log10(lend),np.log10(lstart),npoints)

for alpha in alphas:

simpLasCoefs, Yhat = simpleLasso(X,y,alpha,True)

# print "alpha = ", alpha

# print simpLasCoefs

# print Yhat

simpLasCoefsCR, YhatCR = simpleLasso(X_cr,y_cr,alpha,False)

# print simpLasCoefsCR

# print YhatCR

################------ Exercice 2.4 ------###############################

print "Ex. 2.4: Figure"

#_, theta_lasso, _ =lasso_path(np.array(X), np.array(y), alphas=alphas, fit_intercept=True, return_models=False)

_, theta_lasso_CR, _ =lasso_path(np.array(X_cr), np.array(y_cr), alphas=alphas, fit_intercept=False, return_models=False)

# plot lasso path

# fig1=plt.figure(figsize=(12,8))

# plt.title("Chemin du Lasso: "+ r"$p={0}, n={1} $".format(nfeatures,nsamples),fontsize = 16)

# ax1 = fig1.add_subplot(111)

# ax1.plot(alphas,np.transpose(theta_lasso),linewidth=3)

# ax1.set_xscale('log')

# ax1.set_xlabel(r"$\lambda$")

# ax1.set_ylabel("Amplitude des coefficients")

# ax1.set_ylim([-2,0.5])

# ax1.set_xlim([lstart,lend])

# plt.show(block=False)

fig2=plt.figure(figsize=(12,8))

plt.title("Chemin du Lasso vars_CR: "+ r"$p={0}, n={1} $".format(nfeatures,nsamples),fontsize = 16)

ax1 = fig2.add_subplot(111)

ax1.plot(alphas,np.transpose(theta_lasso_CR),linewidth=3)

ax1.set_xscale('log')

ax1.set_xlabel(r"$\lambda$")

ax1.set_ylabel("Amplitude des coefficients")

ax1.set_ylim([-0.5,0.5])

ax1.set_xlim([lstart,lend])

################------ Exercice 2.5b ------###############################

l_indx = lend

nulvector = np.zeros(nfeatures)

theta = nulvector

while ((theta == nulvector).all()):

l_indx = l_indx-0.01

theta,_ = simpleLasso(X_cr,y_cr,l_indx,False)

#print l_indx," -- ",theta

print "2.5.b) Lambda_0: Plus petit lambda annulant tous les coefs (à moins de 0.01): ", l_indx

# Test de 25 valeurs ll < Lambda_0

print ("\n Test de %i valeurs" % 25)

for lambdapar in np.linspace(l_indx,lstart,25):

coefs,_ = simpleLasso(X_cr,y_cr,lambdapar,False)

print "Lambda: ", lambdapar

print " Coefs: ", coefs

# plt.show(block=False)

# raise SystemExit

print ("\n 2.5.b) CV type leave-one-out, echantillon taille %i" % nsamples)

import sklearn.cross_validation

loo = sklearn.cross_validation.LeaveOneOut(nsamples)

clf = LassoCV(alphas=alphas,fit_intercept=False,normalize=False,cv=loo)

clf.fit(X_cr, y_cr)

coef_lasso=clf.coef_

print "2.5.b) Lambda déterminé par LassoCV:", clf.alpha_

# plot lasso path with CV choice

ax1.axvline(clf.alpha_, color='K',linestyle='-', linewidth= 3)

plt.annotate('CV',

xy=(1.1*clf.alpha_,0.2), xycoords='data',

xytext=(0, 0), textcoords='offset points', fontsize=18)

plt.show(block=False)

filename="lassoCV"

image_name=dirname+filename+imageformat

fig2.savefig(image_name)

################------ Exercice 2.5c ------###############################

xnew = [6,0.3,0.2,6,0.053,25,149,0.9934,3.24,0.35,10]

scorenew = clf.predict(xnew)

print "2.5.c) Prédiction de score pour xnew = \

[6,0.3,0.2,6,0.053,25,149,0.9934,3.24,0.35,10]: ", scorenew

################------ Exercice 2.5d ------###############################

mymodel = linear_model.LinearRegression(fit_intercept=False)

mymodel.fit(X_cr,y_cr)

print "coefs des MCOs"

#print ([mymodel.coef_ , mymodel.intercept_])

print (mymodel.coef_)

theta_LR = mymodel.coef_

# mymodel = linear_model.LinearRegression(fit_intercept=True)

# mymodel.fit(X,y)

# print "coefs des MCOs donnees non CR et ordonnee a l'origine"

# print ([mymodel.coef_ , mymodel.intercept_])

# theta_LR = mymodel.coef_

## Somme des carres des erreurs

## Modele lineaire

ywsfit=mymodel.predict(X_cr)

yws=(y_cr - np.mean(y_cr))/np.std(y_cr)

MSE=(1.0/nsamples)*np.sum((yws-ywsfit)**2)

print "Somme des carrés des erreurs pour la régression linéaire :", MSE

## Somme des carres des erreurs

## Lasso

# clf = LassoCV(alphas=alphas,fit_intercept=False,normalize=False,cv=loo)

# clf.fit(X_cr, y_cr)

print "coefs du Lasso"

print clf.coef_

ywsfit=clf.predict(X_cr)

yws = y_cr

MSE = (1.0/nsamples)*np.sum((yws-ywsfit)**2)

print "Somme des carrés des erreurs pour la régression Lasso :", MSE