with np.load("TINY_MNIST.npz") as data:
x, t = data["x"], data["t"]
x_eval, t_eval = data["x_eval"], data["t_eval"]
xed = np.shape(x_eval)
nl = xed[0]
nc = xed[1]
N_array = [ 5, 50, 100, 200, 400, 800]
N_array_d = np.shape(N_array)
n = N_array_d[0]
Err_Nr = np.zeros(n)
l = 0
for nepochs in N_array:
t_eval_p = np.zeros(nl)
for i in range(0, nl):
t_eval_p[i] = k_NN(x, t, x_eval[i],1, nepochs)
for i in range(0, nl):
if t_eval_p[i] != t_eval[i]:
Err_Nr[l] = Err_Nr[l] + 1
l = l+1
plt.plot(N_array, Err_Nr)
plt.show()
print("T size\t\tErr_Nr")
for i in range (0,n):
print_text = '%d %s %d' % (N_array[i], "\t\t", Err_Nr[i])
print(print_text)
# DRZEWA DECYZYJNE
start_tree = timer()
score_dtree, error_dtree = dtree.dtree(train_inputs, test_inputs,
train_classes, test_classes)
end_tree = timer()
# NAIWNY BAYES
start_bayes = timer()
score_nbayes, error_nbayes = bayes.bayes(train_inputs, test_inputs,
train_classes, test_classes)
end_bayes = timer()
# k-NN k = 3
start_knn3 = timer()
score_knn3, error_knn3 = k_NN.k_NN(3, train_inputs, test_inputs, train_classes,
test_classes)
end_knn3 = timer()
# neural network
start_nn = timer()
score_nn, error_nn = nn.nn(df, train_inputs, test_inputs, train_classes,
test_classes)
end_nn = timer()
# SVM
start_svm = timer()
score_svm, error_svm = svm.svm_f(train_inputs, test_inputs, train_classes,
test_classes)
end_svm = timer()
# k-NN k = 5
x, t = data["x"], data["t"]
x_eval, t_eval = data["x_eval"], data["t_eval"]
xed = np.shape(x_eval)
nl = xed[0]
nc = xed[1]
N = 800
k_array = [1, 3, 5, 7, 21, 101, 401]
k_array_d = np.shape(k_array)
n = k_array_d[0]
Err_Nr = np.zeros(n)
l = 0
for k in k_array:
t_eval_p = np.zeros(nl)
for i in range(0, nl):
t_eval_p[i] = k_NN(x, t, x_eval[i], k, N)
for i in range(0, nl):
if t_eval_p[i] != t_eval[i]:
Err_Nr[l] = Err_Nr[l] + 1
l = l+1
plt.plot(k_array, Err_Nr)
plt.show()
print("k\t\tErr_Nr")
for i in range (0,n):
print_text = '%d %s %d' % (k_array[i], "\t\t", Err_Nr[i])
print(print_text)
