Tensorflow 多分类问题

import requests
import numpy as np

r = requests.get('http://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data')

with open('iris.data', 'w') as f:
    f.write(r.text)

import pandas as pd

data = pd.read_csv('iris.data', names =['e_cd', 'e_kd', 'b_cd', 'b_kd', 'cat'])

data.head(5)

data.cat.unique()

array(['Iris-setosa', 'Iris-versicolor', 'Iris-virginica'], dtype=object)

data['c1'] = np.array(data['cat'] == 'Iris-setosa').astype(np.float32)
data['c2'] = np.array(data['cat'] == 'Iris-versicolor').astype(np.float32)
data['c3'] = np.array(data['cat'] == 'Iris-virginica').astype(np.float32)

target = np.stack([data.c1.values, data.c2.values, data.c3.values]).T

shuju = np.stack([data.e_cd.values, data.e_kd.values, data.b_cd.values, data.b_kd.values]).T

np.shape(shuju), np.shape(target)

((150, 4), (150, 3))

import tensorflow as tf

/anaconda3/envs/py35/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: compiletime version 3.6 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.5
  return f(*args, **kwds)
/anaconda3/envs/py35/lib/python3.5/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters

x = tf.placeholder("float", shape=[None, 4])
y = tf.placeholder("float", shape=[None, 3])

weight = tf.Variable(tf.truncated_normal([4,3]))
bias = tf.Variable(tf.truncated_normal([3]))

combine_input = tf.matmul(x, weight) + bias

pred = tf.nn.softmax(combine_input)

y.get_shape(), pred.get_shape()

(TensorShape([Dimension(None), Dimension(3)]),
 TensorShape([Dimension(None), Dimension(3)]))

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y, logits=combine_input))

WARNING:tensorflow:From <ipython-input-18-6b11325178b9>:1: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.
Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow
into the labels input on backprop by default.

See tf.nn.softmax_cross_entropy_with_logits_v2.

correct_pred = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))

train_step = tf.train.AdamOptimizer(0.0005).minimize(loss)

sess = tf.Session()
sess.run(tf.global_variables_initializer())

for i in range(10000):
    index = np.random.permutation(len(target))
    shuju =shuju[index]
    target = target[index]
    sess.run(train_step, feed_dict={x:shuju, y:target})
    if i%1000 == 0:
        print(sess.run((loss, accuracy), feed_dict={x:shuju, y:target}))

(0.48896936, 0.74)
(0.3105031, 0.91333336)
(0.23715515, 0.9533333)
(0.18639529, 0.96666664)
(0.15062416, 0.96666664)
(0.12501644, 0.98)
(0.10630642, 0.98)
(0.09238311, 0.98)
(0.08186688, 0.98)
(0.07382768, 0.98)

print(sess.run(weight))

[[ 1.319767    1.5383282  -1.6495701 ]
 [ 4.432047    0.22605467 -3.0211918 ]
 [-4.785147   -1.7326641   3.8946218 ]
 [-4.521448   -2.30758     3.069917  ]]

print(sess.run(bias))

[ 1.521937  2.874068 -4.503109]