#####################################################
# #
# test part for testing correctness of #
# Perceptron class #
# #
#####################################################
if __name__ == '__main__':
# small example for testing perceptron class
import matplotlib.pyplot as plt
from read_iris import read_iris_dataset
# read train and test from iris dataset
trainX, validX, testX, trainY, validY, testY = read_iris_dataset(
rand_state=0, train_ratio=0.60, valid_ratio=0.20, test_ratio=0.20)
#print(trainX.shape[0], validX.shape[0], testX.shape[0], trainY.shape[0], validY.shape[0], testY.shape[0])
# for classifying Iris setosa
#trainY = np.where(trainY == 'Iris-setosa', 1, 0)
#testY = np.where(testY == 'Iris-setosa', 1, 0)
# for classifying Iris-versicolor
trainY = np.where(trainY == 'Iris-versicolor', 1, 0)
validY = np.where(validY == 'Iris-versicolor', 1, 0)
testY = np.where(testY == 'Iris-versicolor', 1, 0)
# for classifying Iris-versicolor
#trainY = np.where(trainY == 'Iris-virginica', 1, 0)
#testY = np.where(testY == 'Iris-virginica', 1, 0)
# write in file
file = open('find_good_adalines.txt', 'w')
ratio_of_train_set = 0.7
ratio_of_valid_set = 0.15
ratio_of_test_set = 0.15
min_test_error = 10000.0
best_learning = np.inf
best_random_state = np.inf
for random_state in range(0, 8):
# read train and test from iris dataset
trainX, validX, testX, trainY, validY, testY = read_iris_dataset(rand_state=random_state,
train_ratio=ratio_of_train_set,
valid_ratio=ratio_of_valid_set,
test_ratio=1 - (ratio_of_train_set + ratio_of_valid_set),
order=order)
# print(trainX, testX, trainY, testY)
# standard scaler
scaler = preprocessing.StandardScaler().fit(trainX)
trainX = scaler.transform(trainX)
validX = scaler.transform(validX)
testX = scaler.transform(testX)
# map labels name
map = {'Iris-setosa': 1, 'Iris-versicolor': 2, 'Iris-virginica': 3}
temp = np.zeros(trainY.shape, dtype=int)
for idx, label in enumerate(trainY):
temp[idx] = map[trainY[idx][0]]
# test part for testing correctness of #
# k_fold function #
# #
#####################################################
if __name__ == '__main__':
# small example for testing k-fold function
import numpy as np
import matplotlib.pyplot as plt
from read_iris import read_iris_dataset
from perceptron_one_vs_all import perceptron_one_vs_all
# read train and test from iris dataset
trainX, testX, trainY, testY = read_iris_dataset(rand_state=0,
train_ratio=0.7,
test_ratio=0.15)
#print(trainX, testX, trainY, testY)
#map labels name
map = {'Iris-setosa': 1, 'Iris-versicolor': 2, 'Iris-virginica': 3}
temp = np.zeros(trainY.shape, dtype=int)
for idx, label in enumerate(trainY):
temp[idx] = map[trainY[idx][0]]
trainY = np.copy(temp)
temp = np.zeros(testY.shape, dtype=int)
for idx, label in enumerate(testY):
temp[idx] = map[testY[idx][0]]
testY = np.copy(temp)