import fttlutils
DATA_DIR = "../data/files"
MODEL_DIR = os.path.join(DATA_DIR, "models")
NUM_EPOCHS = 50
BATCH_SIZE = 32
# data
X = np.loadtxt(os.path.join(DATA_DIR, "images-X.txt"), delimiter=",")
y = np.loadtxt(os.path.join(DATA_DIR, "images-y.txt"), delimiter=",",
dtype=np.int)
Y = np_utils.to_categorical(y, nb_classes=5)
np.random.seed(42)
Xtrain, Xtest, Ytrain, Ytest = fttlutils.train_test_split(
X, Y, test_size=0.3, random_state=42)
print(Xtrain.shape, Xtest.shape, Ytrain.shape, Ytest.shape)
# model
# input: (None, 25088)
imgvecs = Input(shape=(Xtrain.shape[1],), dtype="float32")
# hidden layer: (None, 256)
fc1 = Dense(256,
activation="relu",
init="he_uniform",
name="dl1fc1")(imgvecs)
fc1 = BatchNormalization()(fc1)
fc1 = Dropout(0.5)(fc1)
# output layer: (None, 5)
predictions = Dense(5, activation="softmax", name="dl1preds")(fc1)
import numpy as np
import os
import fttlutils
##################### main ######################
DATA_DIR = "../data/files"
MODEL_DIR = os.path.join(DATA_DIR, "models")
# data
X = np.loadtxt(os.path.join(DATA_DIR, "images-X.txt"), delimiter=",")
y = np.loadtxt(os.path.join(DATA_DIR, "images-y.txt"),
delimiter=",",
dtype=np.int)
Xtrain, Xtest, ytrain, ytest = fttlutils.train_test_split(X,
y,
test_size=0.3,
random_state=42)
print(Xtrain.shape, Xtest.shape, ytrain.shape, ytest.shape)
# model
clf = LogisticRegression()
clf.fit(Xtrain, ytrain)
ytest_ = clf.predict(Xtest)
fttlutils.print_stats(ytest, ytest_, "LR Model")
with open(os.path.join(MODEL_DIR, "lr-model.pkl"), "wb") as fmodel:
pickle.dump(clf, fmodel)