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

"""

Spyder Editor

This is a temporary script file.

"""

import numpy as np

import matplotlib.pyplot as plt

import arff

import skfuzzy as fuzz

import pandas as pd

from sklearn.tree import DecisionTreeClassifier

from sklearn.cross_validation import cross_val_score

def plot_res(x1,x2):

ax.set_title('machine learning')

#number = 100

#x = np.random.sample((2,number))

#t = np.random.randint(1,4,number)

#%% open data

data_ps = arff.load(open("c:/Luna/Work/python/data/data_PS/ps_c7_a16_all.arff") )

x = [xx[:-1] for xx in data_ps.get('data')]

t = [xx[-1] for xx in data_ps.get('data')]

#data_ps=pd.DataFrame(x,index = t,columns = [d[0] for d in data.get('attributes')][:-1])

classes={}

xclasses={}

i = 0

for ti in t:

if (classes.get(ti) is None):

classes[ti]=1

xclasses[ti]=[x[i]]

else:

classes[ti]=classes[ti]+1

xclasses[ti].append(x[i])

i=i+1

features_num = len(x[0])

#%% create fuzzy rules

uni_min = np.array([2000000.0 for c in np.arange(float(features_num))])

uni_max = np.array([-1 for c in np.arange(float(features_num))])

feature = [[np.nan] for f in range(0, features_num)]

for key in xclasses.keys():

for f in range(0,features_num):

ff = [x[f] for x in xclasses.get(key)]

[f_min,f_max] = [min(ff),max(ff)]

feature[f] = [f_min,f_max] TERM

uni_min[f] = min([f_min,uni_min[f]])

uni_max[f] = max([f_max,uni_max[f]])

uni_step = (uni_max - uni_min) / 1000

mfs = [[np.nan] for f in range(0, features_num)]

mfs_uni = [[np.nan] for f in range(0, features_num)]

for lingvo in range(0, features_num):

mfs_uni[lingvo] = np.arange(uni_min[lingvo],uni_max[lingvo],uni_step[lingvo])

term = 0

for term_key in xclasses.key():

mfs[lingvo][term] = fuzz.trimf(mfs_uni[lingvp],[feature[x][0],feature[x][0]+(feature[x][1]-feature[x][0])/2,feature[x][1]])

term=term+1

#mf1_universum=np.arange(universum_min,universum_max + universum_step,universum_step)

#mf_feature1 = fuzz.trimf(mf1_universum,[min(feature1),min(feature1)+((max(feature1)-min(feature1))/2),max(feature1)])

#mf_feature2 = fuzz.trimf(mf1_universum,[min(feature2),min(feature2)+((max(feature2)-min(feature2))/2),max(feature2)])

#%% visualization

fig, ax = plt.subplots()

x=0

plt.plot( mfs_uni[x],mfs,'r',mf1_universum,mf_feature2,'g')

#%% something next

#clf= DecisionTreeClassifier(random_state=0)

#cross_val_score(clf, x, t, cv=10)

#plot_res(1,2)

#plt.show()

#f = open("c:/Luna/Work/python/tmp.txt",'rb')

#f = Io.StringIO("Hello")

#data.get('data')

#x = [xx[1:3] for xx in data.get('data')]

#t = [xx[15] for xx in data.get('data')]

#

#data_wine = pd.read_csv("c:/Luna/Work/python/data/wine.data")

#A =data_wine.loc[(data_wine["1"] > 13) & (data_wine["2"] > 1),["1","2"]]

#def num_missing(x):

# return sum(x*x)

#print("Missing values per column:")

#print(data_wine.apply(num_missing, axis=1).head())

#print(data_wine.apply(num_missing, axis=0).head())

#r=data_wine.apply(num_missing, axis=1).head()