MNL(使用自己的数据集)

1. 导入包

import pandas as pd
import numpy as np
from numpy import mean
from numpy import std
from sklearn.datasets import make_classification
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import RepeatedStratifiedKFold
from sklearn.linear_model import LogisticRegression

2. 导入自己的数据

data_wide = pd.read_csv("./data/mode_wide.csv", index_col=0) #index_col=0 第一行为列名
data_wide

	choice	cost.car	cost.carpool	cost.bus	cost.rail	time.car	time.carpool	time.bus	time.rail
1	car	1.507010	2.335612	1.800512	2.358920	18.503200	26.338233	20.867794	30.033469
2	rail	6.056998	2.896919	2.237128	1.855450	31.311107	34.256956	67.181889	60.293126
3	car	5.794677	2.137454	2.576385	2.747479	22.547429	23.255171	63.309057	49.171643
4	car	1.869144	2.572427	1.903518	2.268276	26.090282	29.896023	19.752704	13.472675
5	car	2.498952	1.722010	2.686000	2.973866	4.699140	12.414084	43.092039	39.743252
...	...	...	...	...	...	...	...	...	...
449	rail	6.990901	0.515137	2.066044	2.171174	48.022792	44.501577	27.271918	18.966319
450	car	4.591647	2.891148	1.900379	1.794407	29.444192	33.727087	66.117345	39.842459
451	car	3.236237	1.206815	1.754674	2.023671	16.349017	18.975074	23.387729	43.298276
452	bus	6.932740	1.171861	2.461495	2.612489	65.420641	60.481668	52.404315	48.370662
453	carpool	6.531509	1.408171	2.214791	1.856338	59.566073	55.141406	67.815635	73.447286

453 rows × 9 columns

2. 处理数据

y= 1(选car)；

y = 2 (carpool);

y = 3 (rail);

y = 4 (bus);

def choice_to_y(choice):
    if choice == 'car':
        return 1
    elif choice == 'carpool':
        return 2
    elif choice == 'rail':
        return 3
    else:
        return 4

data_wide['y'] = data_wide['choice'].map(choice_to_y)
data_wide

	choice	cost.car	cost.carpool	cost.bus	cost.rail	time.car	time.carpool	time.bus	time.rail	y
1	car	1.507010	2.335612	1.800512	2.358920	18.503200	26.338233	20.867794	30.033469	1
2	rail	6.056998	2.896919	2.237128	1.855450	31.311107	34.256956	67.181889	60.293126	3
3	car	5.794677	2.137454	2.576385	2.747479	22.547429	23.255171	63.309057	49.171643	1
4	car	1.869144	2.572427	1.903518	2.268276	26.090282	29.896023	19.752704	13.472675	1
5	car	2.498952	1.722010	2.686000	2.973866	4.699140	12.414084	43.092039	39.743252	1
...	...	...	...	...	...	...	...	...	...	...
449	rail	6.990901	0.515137	2.066044	2.171174	48.022792	44.501577	27.271918	18.966319	3
450	car	4.591647	2.891148	1.900379	1.794407	29.444192	33.727087	66.117345	39.842459	1
451	car	3.236237	1.206815	1.754674	2.023671	16.349017	18.975074	23.387729	43.298276	1
452	bus	6.932740	1.171861	2.461495	2.612489	65.420641	60.481668	52.404315	48.370662	4
453	carpool	6.531509	1.408171	2.214791	1.856338	59.566073	55.141406	67.815635	73.447286	2

453 rows × 10 columns

3. 确定自变量X和因变量y

data_wide.columns

Index(['choice', 'cost.car', 'cost.carpool', 'cost.bus', 'cost.rail',
       'time.car', 'time.carpool', 'time.bus', 'time.rail', 'y'],
      dtype='object')

X = data_wide[['cost.car', 'cost.carpool', 'cost.bus', 'cost.rail','time.car', 'time.carpool', 'time.bus', 'time.rail']]
y = data_wide['y']

4. 配置Logit模型并评估

model = LogisticRegression(multi_class='multinomial', solver='lbfgs')

# define the model evaluation procedure （定义模型评估程序） n_splits 就是K-flods中的K值；n_repeats是交叉验证的次数
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=3, random_state=1)
# evaluate the model and collect the scores （评估模型并收集分数）
n_scores = cross_val_score(model, X, y, scoring='accuracy', cv=cv, n_jobs=-1)
# report the model performance 
print('Mean Accuracy: %.3f (%.3f)' % (mean(n_scores), std(n_scores)))  

Mean Accuracy: 0.665 (0.061)

5. 拟合

model.fit(X, y)

D:\ANACONDA\lib\site-packages\sklearn\linear_model\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,





LogisticRegression(multi_class='multinomial')

6. 设置一个新的数据，预测结果

#生成一组新数据
new_data = np.random.rand(8)
new_data

array([0.11880174, 0.16505872, 0.14297278, 0.50355392, 0.87629855,
       0.91189688, 0.57073101, 0.19178997])

#预测
#预测新数据的分布概率
yhat = model.predict_proba([new_data])

#输出预测结果
print('Predicted Probabilities: %s' % yhat[0])

Predicted Probabilities: [0.3749058  0.20228137 0.20380141 0.21901142]


D:\ANACONDA\lib\site-packages\sklearn\base.py:451: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names
  "X does not have valid feature names, but"

已经可以了解如何使用自己的数据进行多元logit回归的一个思路；

上面的警告是出现了无效的特征名（列名不是正确的格式）