'''
学习曲线：模型性能 = f(训练集大小)
学习曲线所需API：
_, train_scores, test_scores = ms.learning_curve(
model, # 模型
输入集, 输出集,
[0.9, 0.8, 0.7], # 训练集大小序列
cv=5 # 折叠数
)
案例：在小汽车评级案例中使用学习曲线选择训练集大小最优参数。
'''
import numpy as np
import matplotlib.pyplot as mp
import sklearn.preprocessing as sp
import sklearn.ensemble as se
import sklearn.model_selection as ms
import sklearn.metrics as sm
import warnings
warnings.filterwarnings('ignore')
data =
[]
with open('./ml_data/car.txt',
'r') as f:
for line
in f.readlines():
sample = line[:-1].split(
',')
data.append(sample)
data =
np.array(data)
# print(data.shape)
# 整理好每一列的标签编码器encoders
# 整理好训练输入集与输出集
data =
data.T
# print(data.shape)
encoders =
[]
train_x, train_y =
[], []
for row
in range(len(data)):
encoder =
sp.LabelEncoder()
if row < len(data) - 1:
# 不是最后列
train_x.append(encoder.fit_transform(data[row]))
else:
# 是最后一列，作为输出集
train_y =
encoder.fit_transform(data[row])
encoders.append(encoder)
train_x =
np.array(train_x).T
# 训练随机森林分类器
model = se.RandomForestClassifier(max_depth=6, n_estimators=150, random_state=7
)
# 绘制学习曲线
train_sizes = np.linspace(0.1, 1, 10
)
_, train_scores, test_scores = ms.learning_curve(model, train_x, train_y, train_sizes=train_sizes, cv=5
)
print(test_scores)
print(np.mean(test_scores,axis=1
))
# 训练之前进行交叉验证
cv = ms.cross_val_score(model, train_x, train_y, cv=4, scoring=
'f1_weighted')
print(cv.mean())
model.fit(train_x, train_y)
# 自定义测试集，预测小汽车的等级
# 保证每个特征使用的标签编码器与训练时使用的标签编码器匹配
data =
[
['high',
'med',
'5more',
'4',
'big',
'low',
'unacc'],
['high',
'high',
'4',
'4',
'med',
'med',
'acc'],
['low',
'low',
'2',
'4',
'small',
'high',
'good'],
['low',
'med',
'3',
'4',
'med',
'high',
'vgood']]
data =
np.array(data).T
test_x, test_y =
[], []
for row
in range(len(data)):
encoder = encoders[row]
# 每列对应的标签编码器
if row < len(data) - 1
:
test_x.append(encoder.transform(data[row])) # 这里需要训练了，直接转换
else:
test_y =
encoder.transform(data[row])
test_x =
np.array(test_x).T
pred_test_y =
model.predict(test_x)
print(pred_test_y)
pred_test_y = encoders[-1
].inverse_transform(pred_test_y)
test_y = encoders[-1
].inverse_transform(test_y)
print(pred_test_y)
print(test_y)
# 画图显示学习曲线
mp.figure(
'Learning Curve', facecolor=
'lightgray')
mp.title('Learning Curve')
mp.xlabel('train size')
mp.ylabel('f1 score')
mp.grid(linestyle=
":")
mp.plot(train_sizes, np.mean(test_scores, axis=1), label=
'Learning Curve')
mp.legend()
mp.show()
输出结果：
[[0.69942197 0.69942197 0.69942197 0.69942197 0.70348837
]
[0.67630058 0.79768786 0.69942197 0.71965318 0.70348837
]
[0.66184971 0.70231214 0.75433526 0.74855491 0.70348837
]
[0.71098266 0.78323699 0.74277457 0.73988439 0.7005814
]
[0.71387283 0.71965318 0.5982659 0.74277457 0.74127907
]
[0.71387283 0.76878613 0.70809249 0.74855491 0.73837209
]
[0.71387283 0.7716763 0.72254335 0.82080925 0.75872093
]
[0.71387283 0.76878613 0.72254335 0.83526012 0.75872093
]
[0.71387283 0.7716763 0.73121387 0.83526012 0.76744186
]
[0.73121387 0.76878613 0.72254335 0.8583815 0.86046512
]]
[0.70023525 0.71931039 0.71410808 0.735492 0.70316911 0.73553569
0.75752453 0.75983667 0.763893 0.78827799
]
0.7477732938195376
[2 0 0 3
]
['unacc' 'acc' 'acc' 'vgood']
['unacc' 'acc' 'good' 'vgood']
转载于:https://www.cnblogs.com/yuxiangyang/p/11194207.html
