File size: 3,754 Bytes
8a79f2e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 |
import pandas as pd
from stochatreat import stochatreat
import random
from functools import reduce
random.seed(13984759)
"""Assigns treatments in prep for the assignment Survey. If there is a used_cl, the assign_treatment function will use
ActualUse data from the used CL
- Treatment Assignment will use data from the old_phase
- New treatment assignment will take effect in the new_phase"""
class Treatment():
def __init__(self,seed):
self.i = seed
def prepare_strat(self, df,continuous_strat,discrete_strat):
#discretize continuous strat
for var in continuous_strat:
df[var] = df[var].astype(float)
median = df[var].median()
df.loc[df[var] >= median,var+"Strat"] = f"{var}High"
df.loc[df[var] < median, var + "Strat"] = f"{var}Low"
#If var is missing a strat value, put in low
df.loc[df[var].isnull(), var + "Strat"] = f"{var}Low"
#label discrete strat
for var in discrete_strat:
df[var+"Strat"] = df[var].astype(str).apply(lambda x: var+x)
#compose strat var
strat_vars = [x+"Strat" for x in discrete_strat+continuous_strat]
df["Stratifier"] = df[strat_vars].values.tolist()
df["Stratifier"] = df["Stratifier"].apply(lambda x: 'X'.join(x))
return df
""" Used to assign randomly assign treatment to a subset of the data. The varname should already be in the data set. The function only modifies the values for the subset
Input:
- df: the full dataframe
- subset_var: the categorical vairable used to subset
- subset_val: the value of subset_var we will keep
- inputs to _assign_treat_var
Outpu:
- the full df with the varname randomly filled"""
def subset_treat_var_wrapper(self,df, subset_var, subset_val, rand_dict: dict, stratum_cols: list, varname):
r_varname = "Randomized" + varname
df_dl = self.assign_treat_var(df=df.loc[df[subset_var] == subset_val],
rand_dict=rand_dict,
stratum_cols=stratum_cols,
varname=r_varname)
df = df.merge(df_dl[["AppCode", r_varname]], how='left', on="AppCode")
df.loc[df[subset_var] == subset_val, varname] = df.loc[
df[subset_var] == subset_val, r_varname]
df = df.drop(columns=[r_varname])
return df
def assign_treat_var(self, df: pd.DataFrame, rand_dict: dict, stratum_cols: list, varname: str):
treats = stochatreat(data=df,
stratum_cols=stratum_cols,
treats=len(rand_dict.keys()),
probs=list(rand_dict.values()),
idx_col='AppCode',
random_state= self.i,
misfit_strategy='stratum'
)
self.i = self.i * 2
raw_treat_values = range(len(rand_dict.keys()))
treat_labels = list(rand_dict.keys())
translate_dict = dict(zip(raw_treat_values, treat_labels))
treats[varname] = treats["treat"].apply(lambda x: translate_dict[x])
treat_tab = pd.crosstab(treats["stratum_id"], treats[varname], margins=True).reset_index()
df = df.merge(treats[["AppCode", varname]], on="AppCode")
# Assert that the number of strata equal the number of rows when group the treatment df by strata
value_len = [len(df[x].unique()) for x in stratum_cols]
number_of_strata = reduce(lambda x, y: x * y, value_len)
assert len(treats["stratum_id"].unique()) == number_of_strata
return df
|