from NN import NN
from dataset.hosing import hosing
if __name__ == "__main__":
# load the train data
X, Y = hosing.load()
# data partition
test_X = X[-50:]
test_Y = Y[-50:]
train_X = X[:-50]
train_Y = Y[:-50]
# create nn model
nn = NN.NeuralNetwork([13, 10, 1], 'ReLu')
# init
nn.init('other')
# fit
nn.fit(train_X,
train_Y,
epochs=10000,
lr=0.0001,
normalization=True,
batch=16,
loss='mse')
# predict
right = 0
all = 0
mae = 0
for i in zip(test_X, test_Y):
result = nn.predict(i[0])
from NN import NN
from dataset.mnist import mnist
import numpy as np
if __name__ == "__main__":
# load the data
train_images, train_labels = mnist.load_train()
test_images, test_labels = mnist.load_test()
# crate nn model
nn = NN.NeuralNetwork([784, 100, 10], 'sig')
# init
nn.init('normal')
# fit
nn.fit(train_images,
train_labels,
epochs=5000,
lr=0.1,
loss='mae',
batch=1)
# predict
right = 0
all = 0
for i in zip(test_images, test_labels):
result = nn.predict(i[0])
if np.argmax(result) == np.argmax(i[1]):
right += 1
all += 1
print("precision", right / all)
