#Author: Kairat Tilegen
#Email: kairat.tilegen@is.sdu.edu.kz

import csv # is used to read csv file
import shlex # is used to split avoiding unecessary symbols'_'
import copy # is used to copy whole array to make new Dataset
import numpy as np
from sklearn.neighbors import KNeighborsClassifier
from sklearn.cross_validation import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.tree import DecisionTreeClassifier
from sklearn.linear_model import LogisticRegression, LinearRegression
from sklearn import linear_model
import matplotlib.pyplot as plt
from sklearn import metrics
import pandas as pd
from sklearn.naive_bayes import GaussianNB
import random # For colors

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

#Working with Data Converting Words to Numeric using Dictionary
#Reading Banking.csv file and implementing it
filename = '/users/kairat/Desktop/bank.csv'

 #job
f2 = {"admin.":0,"unknown":1,"unemployed":2,"management":3,"housemaid":4,"entrepreneur":5,"student":6,"blue-collar":7,"self-employed":8,"retired":9,"technician":10,"services":11}
#marital
f3 = {"married":0,"divorced":1,"single":2}
#education
f4 ={ "unknown":0,"secondary":1,"primary":2,"tertiary":3}
    #default
f5 = { 'no':1,'yes':0}
    #housing
f7 = {'no':1,'yes':0}
    #loan
f8 = {'no':1,'yes':0}
    #contact
f9 = {'unknown':0,'cellular':1,'telephone':2}
    #month
f11 = {"jan":0, "feb":1, "mar":2, "apr":3,"may":4,"jun":5,"jul":6,"aug":7,"sep":8,"oct":9, "nov":10, "dec":11}
    #poutcome
f16 = { "unknown":0,"other":1,"failure":2,"success":3}
    #y
f17 = {"yes":0,"no":1 }
with open(filename, 'rb') as f:
    reader = csv.reader(f)#Read via csv built-in function
    dataset=[]
    data=[]
    target=[]
    try:
        for row in reader:
            line=shlex.split(row[0])#Get rid of spaces other symbols..
            dataset.append(line[0].split(';'))
    except csv.Error as e:
        sys.exit('file %s, line %d: %s' % (filename, reader.line_num, e))
    features=[]
    for i in dataset[0]:
        features.append(i)
    j=0
    #TRANSFORMING INFORMATION TO DATA AND TARGET
    for i in range(1,len(dataset)):
        data.append([])
        data[j].append(float(dataset[i][0]))
        data[j].append(float(f2[dataset[i][1]]))
        data[j].append(float(f3[dataset[i][2]]))
        data[j].append(float(f4[dataset[i][3]]))
        data[j].append(float(f5[dataset[i][4]]))
        data[j].append(float(dataset[i][5]))
        data[j].append(float(f7[dataset[i][6]]))
        data[j].append(float(f8[dataset[i][7]]))
        data[j].append(float(f9[dataset[i][8]]))
        data[j].append(float(dataset[i][9]))
        data[j].append(float(f11[dataset[i][10]]))
        data[j].append(float(dataset[i][11]))
        data[j].append(float(dataset[i][12]))
        data[j].append(float(dataset[i][13]))
        data[j].append(float(dataset[i][14]))
        data[j].append(float(f16[dataset[i][15]]))
        target.append(float(f17[dataset[i][16]]))
        j+=1
    newdata=copy.deepcopy(data)

    for i in range(0,j):
         newdata[i].append(target[i])
    # NEWDATA IS NEW DATASET
#######################################################################

# Working with data
x=np.asarray(data)
y=np.asarray(target)
X_train, X_test, Y_train, Y_test = train_test_split(x, y, test_size=0.33,random_state=0)

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


#Getting prepared methods from prediction classes
#Bayes
nb=GaussianNB()
nb.fit(X_train,Y_train)
nbpred=[]
#KNN
knn=KNeighborsClassifier(n_neighbors=3)
knn.fit(X_train,Y_train)
knnpred=[]
#DT
model = DecisionTreeClassifier(min_samples_split=5)
model.fit(X_train, Y_train)
dtpred=[]
#LR
logit = LogisticRegression(C=0.5)
logit.fit(X_train,Y_train)
logitpred=[]
#LRM
regr = linear_model.LogisticRegression()
regr.fit(x, y)
lrm=[]
#########################################
#preparing DATA for making Predictions
datass=list(X_test)# CONVERT FROM NP TO LIST
for i in range(0,len(datass)):
    ndata=datass[i]
    knnpred.append(knn.predict([datass[i]]))
    dtpred.append(model.predict([ndata]))
    nbpred.append(nb.predict([ndata]))
    logitpred.append(logit.predict([ndata]))
    lrm.append(regr.predict([ndata]))

def manipulation():# GIVING INFORMATION ABOUT HOW DATA WAS MANIPULATED
    for i in range(0,len(datass)):
        print "DATA:",ndata
        print "KNN class predicted :",(knn.predict([ndata]))
        print "KNN probability:",(knn.predict_proba([ndata]))
        print "DT class predicted:",(model.predict([ndata]))
        print "DT proba:",(model.predict_proba([ndata]))
        print "NB class:",(nb.predict([ndata]))
        print "NB probability:",nb.predict_proba([ndata])
        print "Logit class :",(logit.predict([ndata]))
        print "Logit probability:",logit.predict_proba([ndata])
        print "LRM class :",(regr.predict([ndata]))
        print "LRM probability:",regr.predict_proba([ndata])
def accuracies():
    print "accuracy KNN Algorithm:",accuracy_score(Y_test, knnpred)
    print "accuracy Data Tree:",accuracy_score(Y_test, dtpred)
    print "accuracy Gaussian Normal:",accuracy_score(Y_test,nbpred)
    print "accuracy Logistic Regression:",accuracy_score(Y_test, logitpred)
    print "accuract LRM",accuracy_score(Y_test,lrm)
def logits():
    print('Coefficients: \n', logit.coef_)
    # Explained variance score: 1 is perfect prediction
    print('Variance: %.2f' % logit.score(X_test, Y_test))

    print pd.DataFrame(zip(features,logit.coef_),columns=['features','EstCoeff'])
    y_pred=logit.predict(X_test)
    print("MSE: %.2f" % (metrics.mean_squared_error(Y_test,y_pred)))
    print("MAE: %.2f" % (metrics.mean_absolute_error(Y_test,y_pred)))
    print("RMSE: %.2f" % (np.sqrt(metrics.mean_squared_error(Y_test,y_pred))))


# TO FIND THE HIGHEST CORRELATION x>0.6
indexes=[]
forplot=[]
data_frame =pd.DataFrame(newdata)
data_frame.head()
data_frame.columns=features
result = data_frame.y
columns=[(data_frame.age),(data_frame.job),(data_frame.marital),(data_frame.education),(data_frame.default),(data_frame.balance),(data_frame.housing),(data_frame.loan),(data_frame.contact),(data_frame.day),(data_frame.month),(data_frame.duration),data_frame.campaign,(data_frame.pdays),(data_frame.previous),(data_frame.poutcome),data_frame.y]
column=[]
row=[]
for i in range(0,len(columns)):
    for j in range(i+1,len(columns)-1):
        if columns[i].corr(columns[j])>0.6:
            forplot.append(columns[i].corr(columns[j]))
            column.append(features[i])
            row.append(features[j])
def corrr():
    for i in range(0,len(columns)):
        for j in range(i+1,len(columns)-1):
            print 'corr btw',features[i],'and',features[j],columns[i].corr(columns[j])

def high_cor():
    print 'WITH HIGH CORRELATIONS', forplot
plt.figure(figsize=(45, 15))
plots = len(forplot)
ax=[]
s=0
f=0

##############################
# Drawing Scatter Plot
def scatter_plot():
    plt.figure(figsize=(45, 15))
    plots = len(forplot)
    ax=[]
    s=0
    f=0
    for i in range(0,plots):
            ax.append(plt.subplot2grid((5,4), (s,f)))
            f+=1
            ax[i].scatter(data_frame[row[i]],data_frame[column[i]],  s=10, c=[random.random(),random.random(),random.random()], marker="o")
            #ax[i].plot(data_frame[row[i]],data_frame[column[i]])
            ax[i].set_ylabel(row[i])
            ax[i].set_xlabel(column[i])
            if (i+1)%4==0:
                s+=1
                f=0
    plt.show()
inp='lylylylalyalyalyal'
while inp!=' ':
    print "1 - Manipulation"
    print "2 - Accuracies"
    print '3 - Show Scatter Plot'
    print '4 - High Corrs'
    print '5 - Logit Implementations'
    print '6 - CORRS between features'
    inp=raw_input('Enter The command: ')
    if inp=='1':
        manipulation()
    elif inp=='2':
        accuracies()
    elif inp=='3':
        scatter_plot()
    elif inp=='4':
        high_cor()
    elif inp=='5':
        logits()
    elif inp=='6':
        corrr()