from sklearn.preprocessing import MinMaxScaler from sklearn.decomposition import PCA from sklearn.feature_selection import SelectKBest from sklearn.utils import shuffle from sklearn.naive_bayes import GaussianNB from sklearn.manifold import LocallyLinearEmbedding from data.preprocess import features_preprocess, features_test_preprocess, labels_preprocess, labels_preprocess_num from data.preprocess_2nd import preprocess_ft_lbls_num from data.preprocess import ft_lbls_num scores = [] embedding = LocallyLinearEmbedding(n_components=10) (features1, labels1) = ft_lbls_num() (features2, labels2) = preprocess_ft_lbls_num() features1 = embedding.fit_transform(features1, labels1) features2 = embedding.fit_transform(features2, labels2) K = 5 cv = KFold(n_splits=K, shuffle=True) features = numpy.concatenate((features1, features2)) labels = numpy.concatenate((labels1, labels2)) clf = svm.SVC(kernel='rbf') for i in range(100):
import keras import sklearn from sklearn import svm from keras.layers import Dense, Input, LeakyReLU from keras.models import Model from sklearn.feature_selection import SelectKBest from sklearn.metrics import accuracy_score from sklearn.model_selection import KFold import numpy from data.preprocess import ft_lbls_num from data.preprocess_2nd import preprocess_ft_lbls_num (features, labels) = ft_lbls_num() labels = numpy.asarray(labels, dtype=numpy.float32) input = Input(shape=(50, )) hd1 = Dense(20)(input) leaky = LeakyReLU()(hd1) output = Dense(50, activation='softmax')(hd1) model = Model(input, output) encoder = Model(input, hd1) model.compile(optimizer='adam', loss='binary_crossentropy') K = 5 cv = KFold(n_splits=K, shuffle=True) scores = [] for train, test in cv.split(features):
