0

我非常陌生，causal inference所以我所做的就是使用库示例尽可能地接近我的潜在答案。我试图使用 causlaml 库来弄清楚它是如何工作的。我将样本分成control group、1000x3、y = target和treatment group与对照组相同的形状。但是在图书馆里面的例子中，形状是不同的。代码如下：

from causalml.inference.meta import LRSRegressor
from causalml.inference.meta import XGBTRegressor, MLPTRegressor
from causalml.inference.meta import BaseXRegressor
from causalml.inference.meta import BaseRRegressor
from xgboost import XGBRegressor
from causalml.dataset import synthetic_data
import numpy as np
import pandas as pd

###Prparing data
df = pd.read_csv('sample_data/configuration_900.out',sep = "\s+",header = None)

df.columns = ['x','y','s','theta','curvature','near']

MIDD = 500

# Function to find middle index
def find_midd(input_list): #middleIndex %2 != 0:
    middleIndex = int(float(len(input_list))) /2
    if middleIndex %2 == 0:
        return middleIndex 
    else:
        return middleIndex - 0.5

excess = df.iloc[:,-1]

middleindex = find_midd(excess)
middleindex = int(middleindex)

new_theta_up = df.iloc[:middleindex,3]
new_cur_up   = df.iloc[:middleindex,4]
new_exc_up   = df.iloc[:middleindex,5]


df_up = {'col1':new_theta_up,'col2':new_cur_up,'col3':new_exc_up}

df_up = pd.DataFrame(df_up)

df_up.columns = ['theta','Curvature','Target']

new_theta_down = df.iloc[middleindex:-1,3]       
new_cur_down   = df.iloc[middleindex:-1,4]
new_exc_down   = df.iloc[middleindex:-1,5]

df_down = {'col1':new_theta_down,'col2':new_cur_down,'col3':new_exc_down}
df_down = pd.DataFrame(df_down)

df_down.columns = ['theta','Curvature','Target']

targ = excess[middleindex]

new_targ = np.repeat(targ,len(df_down))

### Designing the causal model
y = new_targ


X = df_up

y = new_targ

treatment = df_down
### The linear regression estimator

lr = LRSRegressor()
te, lb, ub = lr.estimate_ate(X, treatment, y)
print('Average Treatment Effect (Linear Regression): {:.2f} ({:.2f}, {:.2f})'.format(te[0], lb[0], ub[0]))

它会抛出如下错误：

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-37-c86082e758d7> in <module>()
     68 
     69 lr = LRSRegressor()
---> 70 te, lb, ub = lr.estimate_ate(X, treatment, y)
     71 print('Average Treatment Effect (Linear Regression): {:.2f} ({:.2f}, {:.2f})'.format(te[0], lb[0], ub[0]))
     72 

2 frames
/usr/local/lib/python3.7/dist-packages/causalml/inference/meta/utils.py in check_treatment_vector(treatment, control_name)
     17     if control_name is not None:
     18         assert control_name in treatment, \
---> 19             'Control group level {} not found in treatment vector.'.format(control_name)
     20 
     21 

AssertionError: Control group level 0 not found in treatment vector.
4

0 回答 0