OrthoForest spend days working without result

[juandavidgutier]

Hello everyone,

I am running OrthoForest from DoWhy but it is running for days without result. I get this message:

"2022-01-16 20:57:41.811540: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found
2022-01-16 20:57:41.811571: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine."

Any idea how to solve it?

Here is the dataset:
new_incidence_sd_BC_test2_ml.csv

This is my code:

`

importing required libraries

import os, warnings, random
import dowhy
import econml
from dowhy import CausalModel
import pandas as pd
import numpy as np

data = pd.read_csv("new_incidence_sd_BC_test2_ml.csv")
#data_inf = data_inf.astype({"treatment":'bool'}, copy=False)
data = data.dropna()

#data['treatment'] = data['treatment']*1000000000 #micrograms
data_inf = data.drop(['X2','X3','X4'], axis=1)
data_inf.head()

#Step 1: Modeling the causal mechanism
#model with acumm PM2.5 as treatment, GEI and Temperatura(W) as common causes, Rainfall_accum2s as instrument
model_inf=CausalModel(
data = data_inf,
treatment=['treatment'],
outcome='y',
instruments=['Z1', 'Z2'],
common_causes=['W1', 'W2'],
#common_causes=data_inf[['W1','W2']].to_numpy().reshape(-1, 2),
#effect_modifiers=Xs.split('+'),
effect_modifiers= ['X1'],
graph= "digraph {W1->treatment;W1->y;W2->treatment;W2->y;Z1->treatment;Z2->treatment;W2->W1;treatment->y;X1->y;}"
)

#view model
model_inf.view_model()

#Step 2: Identifying effects
identified_estimand = model_inf.identify_effect(proceed_when_unidentifiable=True)
print(identified_estimand)

#Step 3: Estimation of the effect
#with OrthoForest
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.linear_model import LassoCV
from econml.orf import DMLOrthoForest
from econml.sklearn_extensions.linear_model import WeightedLassoCVWrapper

#step 3: Estimation of the effect
orf_estimate_bd = model_inf.estimate_effect(identified_estimand,
#test_significance=True,
confidence_intervals=True,
method_name="backdoor.econml.orf.DMLOrthoForest",
method_params={
'init_params': {'n_trees':250,
'min_leaf_size':20,
'max_depth':10,
#'bootstrap':True,
'lambda_reg':0.01,
'random_state':123,
'model_Y':GradientBoostingRegressor(),
'model_T': GradientBoostingRegressor(),
'model_T_final':None, #WeightedLassoCVWrapper(cv=3),
'model_Y_final':None,}, #WeightedLassoCVWrapper(cv=3),
'fit_params': {}
})

print(orf_estimate_bd)
`