https://www.kaggle.com/c/bike-sharing-demand/data

bike_cnn.py

import numpy as np
import pandas as pd
import tensorflow as tf

train = pd.read_csv("data_bike/train.csv", parse_dates=["datetime"])

test = train[:2000]
train= train[2000:]

feature=['temp', 'atemp', 'humidity', 'windspeed', 'casual', 'registered']
x_data = train[feature]

y_data = train[['count']]

x_data = np.array(x_data, dtype=np.float32).reshape((-1, 2, 3, 1))
y_data = np.array(y_data, dtype=np.int32)

X = tf.placeholder(tf.float32, [None, 2, 3, 1])
Y = tf.placeholder(tf.int32, [None, 1])
conv1 = tf.layers.conv2d(inputs=X, filters=10, kernel_size=[2, 2],
      padding="same", activation=tf.nn.relu)
conv2 = tf.layers.conv2d(inputs=conv1, filters=15, kernel_size=[2, 2],
      padding="same", activation=tf.nn.relu)
conv3 = tf.layers.conv2d(inputs=conv2, filters=20, kernel_size=[3, 3],
      padding="same")
pool1 = tf.layers.max_pooling2d(inputs=conv3, pool_size=[2, 2], strides=1)

flatten = tf.contrib.layers.flatten(pool1)
fc1 = tf.layers.dense(inputs=flatten, units=30, activation=tf.nn.relu)
fc2 = tf.layers.dense(inputs=fc1, units=20, activation=tf.nn.relu)
fc3 = tf.layers.dense(inputs=fc2, units=10, activation=tf.nn.relu)
fc4 = tf.layers.dense(inputs=fc3, units=1)
loss = tf.losses.mean_squared_error(Y, fc4)
tf.summary.scalar('loss', loss)
optimizer = tf.train.AdamOptimizer().minimize(loss)

init = tf.global_variables_initializer()
summary = tf.summary.merge_all()

sess = tf.Session()
sess.run(init)
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter('ttrain', sess.graph)

for idx in range(10000):
    s, c, _= sess.run([summary, loss, optimizer], feed_dict={
            X: x_data,
            Y: y_data
        })
    train_writer.add_summary(s, global_step=idx)
    
    if (idx+1) % 100 == 0:
      print(idx, c)

feature=['temp', 'atemp', 'humidity', 'windspeed', 'casual', 'registered']
x_test = test[feature]
x_test = np.array(x_test).reshape((-1, 2, 3, 1))

y_test = test[['count']]
y_tesst = np.array(y_test)

print('test predict')
accuracy = sess.run(loss, feed_dict={
        X: x_test,
        Y: y_test
    })

print(1 - accuracy)

bike_dnn.py

# https://www.kaggle.com/c/bike-sharing-demand/data
import numpy as np
import pandas as pd
import tensorflow as tf

train = pd.read_csv("data_bike/train.csv", parse_dates=["datetime"])
test = pd.read_csv("data_bike/test.csv", parse_dates=["datetime"])

label_name = "count"
y_data = train[['count']]
print(y_data.shape)

feature=['temp', 'atemp', 'humidity', 'windspeed', 'casual', 'registered']
x_data = train[feature]
x_data.head()

X = tf.placeholder(tf.float32, [None, 6])
Y = tf.placeholder(tf.int32, [None, 1])

layer1 = tf.layers.dense(inputs=X, units=100, activation=tf.nn.relu)
layer2 =tf.layers.dense(inputs=layer1, units=90, activation=tf.nn.relu)
layer3 =tf.layers.dense(inputs=layer2, units=50, activation=tf.nn.relu)
layer4 =tf.layers.dense(inputs=layer3, units=30, activation=tf.nn.relu)
layer5 =tf.layers.dense(inputs=layer4, units=1)

cost = tf.losses.mean_squared_error(Y, layer5)
tf.summary.scalar('cost', cost)
train = tf.train.AdamOptimizer().minimize(cost)

init = tf.global_variables_initializer()
summary = tf.summary.merge_all()

with tf.Session() as sess:
    sess.run(init)
    merged = tf.summary.merge_all()
    train_writer = tf.summary.FileWriter('train', sess.graph)
    
    for idx in range(10000):
        s, c, _= sess.run([summary, cost, train], feed_dict={
                X: x_data,
                Y: y_data
            })
        train_writer.add_summary(s, global_step=idx)
        
        if idx % 1000 == 0:
            print(idx, c)