使用keras对鸢尾花进行多分类

参考代码：

import keras
import numpy as np
from keras.losses import CategoricalCrossentropy
from keras.optimizers import SGD
from keras.utils import to_categorical
from numpy import random
from sklearn.datasets import load_iris
from sklearn.metrics import confusion_matrix, recall_score

# 加载数据
(data, target) = load_iris(return_X_y=True)

# 数据打乱
indices = list(range(len(target)))
random.shuffle(indices)
data = data[indices]
target = target[indices]

tag_count = len(set(target))

# 构造模型
model = keras.models.Sequential(
    layers=[keras.layers.Dense(512, activation='relu', input_shape=(4,)),
            keras.layers.Dropout(0.2),
            keras.layers.Dense(tag_count, activation='softmax')]
)

print(model.summary())
# Model: "sequential_1"
# _________________________________________________________________
# Layer (type)                 Output Shape              Param #
# =================================================================
# dense_1 (Dense)              (None, 512)               2560
# _________________________________________________________________
# dropout_1 (Dropout)          (None, 512)               0
# _________________________________________________________________
# dense_2 (Dense)              (None, 3)                 1539
# =================================================================
# Total params: 4,099
# Trainable params: 4,099
# Non-trainable params: 0


model.compile(optimizer=SGD(),
              loss=CategoricalCrossentropy(), metrics=['accuracy'])

target = to_categorical(target, tag_count)

# 数据切分为两部分 80% 用于训练，20%用于预测
train_size = int(len(target) * 0.8)

train_x, train_y = data[:train_size], target[:train_size]
test_x, test_y = data[train_size:], target[train_size:]

history = model.fit(train_x, train_y, batch_size=10, epochs=100, validation_split=0.25)

pred_y = model.predict(test_x)

print(pred_y[:3])
# [[5.1133092e-03 6.2675482e-01 3.6813188e-01]
#  [5.4596356e-05 4.3929096e-02 9.5601630e-01]
#  [9.6804839e-01 3.1915616e-02 3.6038000e-05]]

# 根据softmax结果获取类别，即：最大softmax值对应的下标
test_y_pred = [np.argmax(y, axis=None, out=None) for y in pred_y]
test_y_true = [np.argmax(y, axis=None, out=None) for y in test_y]

print(test_y_true)
print(test_y_pred)
# [1, 2, 0, 2, 0, 1, 2, 0, 1, 0, 2, 2, 1, 2, 1, 0, 0, 1, 2, 2, 2, 2, 1, 2, 0, 1, 1, 1, 2, 2]
# [1, 2, 0, 2, 0, 1, 2, 0, 1, 0, 2, 2, 1, 2, 1, 0, 0, 1, 2, 2, 2, 2, 1, 2, 0, 1, 2, 1, 2, 2]


print(confusion_matrix(test_y_true, test_y_pred))
# [[ 7  0  0]
#  [ 0  9  1]
#  [ 0  0 13]]


print(recall_score(test_y_pred, test_y_true, average='micro'))
# 0.9666666666666667