# -*- coding: utf-8 -*-
"""
Created on Fri Apr  5 10:50:47 2019

@author: mattdoe
"""

from data_preprocessor_db import data_storage # validation data
from sklearn.preprocessing import MinMaxScaler
from sklearn import svm
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
from numpy import array
import pickle



# for seperation of data_storage
# Link_ID = list()
Input, Output = list(), list()

# seperate data_storage in Input and Output data
for items in data_storage:
    # Link_ID = items[0] # identifier not needed
    Input.append((float(items[1]), float(items[2]), float(items[3]), float(items[4]), float(items[5]), float(items[6]), float(items[7]), float(items[8]), float(items[9]))) # Input: all characteristics
    Output.append(float(items[10])) # Output: scenario_class 1 to 9

# Input tuple to array
A = array(Input)

# normalise array between 0 and 1
scaler = MinMaxScaler(feature_range=(-1, 1))
scaledA = scaler.fit_transform(A)

# Output tuple to array
B = array(Output)

# split train and test data; ration: 70:30
# shuffle = False: doesn't sort data randomly
# shuffle = True: default: sorts data randomly
A_train, A_test, B_train, B_test = train_test_split(A, B, test_size=0.3, shuffle=True, random_state=40)

# create model
model = svm.SVC(kernel='linear', C = 1.0)

# fit model
model.fit(A_train, B_train)

# get support vectors
# model.support_vectors_

# get indices of support vectors
# model.support_

# get number of support vectors for each class
# model.n_support_

filename = 'ml_svm.sav'
pickle.dump(model, open(filename, 'wb'))

# load the model from disk
loaded_model = pickle.load(open(filename, 'rb'))

# test to all data records
# result = loaded_model.score(A, B)

# test with test data
# score represents the mean accuracy of given test data and labels
result = loaded_model.score(A_test, B_test) # relative
print(result)

# confusion matrix compares true value with predicted value
# true value <--> predicted value
predicted = model.predict(A_test)
tn, fp, fn, tp = confusion_matrix(B_test, predicted, labels=[1, 2, 3, 4, 5, 6, 7, 8, 9]).ravel()

Traceback (most recent call last):

  File "<ipython-input-8-8649dd873bbd>", line 1, in <module>
    runfile('C:/Workspace/Master-Thesis/Programm/MapValidationML/ml_svm.py', wdir='C:/Workspace/Master-Thesis/Programm/MapValidationML')

  File "C:ProgramDataAnaconda3libsite-packagesspyder_kernelscustomizespydercustomize.py", line 827, in runfile
    execfile(filename, namespace)

  File "C:ProgramDataAnaconda3libsite-packagesspyder_kernelscustomizespydercustomize.py", line 110, in execfile
    exec(compile(f.read(), filename, 'exec'), namespace)

  File "C:/Workspace/Master-Thesis/Programm/MapValidationML/ml_svm.py", line 75, in <module>
    tn, fp, fn, tp = confusion_matrix(B_test, predicted, labels=[1, 2, 3, 4, 5, 6, 7, 8, 9]).ravel()

ValueError: too many values to unpack (expected 4)