NSQIP_codes/cpt_baseline.py at master · goldenberg-lab/NSQIP_codes · GitHub

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
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
import numpy as np
import pandas as pd
import os
import dill
from support.support_funs import stopifnot, find_dir_nsqip
from support.naive_bayes import mbatch_NB
from sklearn import metrics
import seaborn as sns


###############################
# ---- STEP 1: LOAD DATA ---- #

dir_NSQIP = find_dir_nsqip()
dir_output = os.path.join(dir_NSQIP,'output')
dir_models = os.path.join(dir_output, 'models')
dir_figures = os.path.join(dir_NSQIP,'figures')
lst_dir = [dir_figures, dir_output, dir_models]
assert all([os.path.exists(fold) for fold in lst_dir])

fn_X = 'X_imputed.csv'
fn_Y = 'y_agg.csv'
dat_X = pd.read_csv(os.path.join(dir_output,fn_X))  #,usecols=['operyr','caseid','cpt']
dat_Y = pd.read_csv(os.path.join(dir_output,fn_Y))
print(dat_X.shape); print(dat_Y.shape)
stopifnot(all(dat_X.caseid == dat_Y.caseid))
u_years = dat_X.operyr.unique()
# !! ENCODE CPT AS CATEGORICAL !! #
dat_X['cpt'] = 'c'+dat_X.cpt.astype(str)
cn_X = list(dat_X.columns[2:])
cn_Y = list(dat_Y.columns[2:])

###############################################
# ---- STEP 2: LEAVE-ONE-YEAR - CPT ONLY ---- #

holder_vv = []
holder_phat = []
for ii, vv in enumerate(cn_Y):
    print('##### ------- Outcome %s (%i of %i) -------- #####' % (vv, ii+1, len(cn_Y)))
    tmp_ii = pd.concat([dat_Y.operyr, dat_Y[vv]==-1],axis=1)
    tmp_ii = tmp_ii.groupby('operyr')[vv].apply(np.sum).reset_index().rename(columns={vv:'n'})
    tmp_years = tmp_ii[tmp_ii.n == 0].operyr.values
    tmp_train_years = tmp_years[tmp_years > tmp_years.min()]
    holder_metrics = []
    holder_score = []
    for yy in tmp_train_years:
        print('Year %i' % (yy))
        idx_train = dat_X.operyr.isin(tmp_years) & (dat_X.operyr < yy)
        idx_test = dat_X.operyr.isin(tmp_years) & (dat_X.operyr == yy)
        Xtrain, Xtest = dat_X.loc[idx_train, ['cpt']].reset_index(drop=True), \
                        dat_X.loc[idx_test, ['cpt']].reset_index(drop=True)
        ytrain, ytest = dat_Y.loc[idx_train,vv].reset_index(drop=True), \
                        dat_Y.loc[idx_test,vv].reset_index(drop=True)
        # --- train model --- #
        mdl_bernoulli = mbatch_NB(method='bernoulli')
        mdl_bernoulli.fit(Xtrain,ytrain.values,mbatch=100000)
        phat_bernoulli = mdl_bernoulli.predict(Xtest)[:,1]
        case_id = dat_X.loc[idx_test, 'caseid']
        holder_score.append(pd.DataFrame({'y':ytest.values,'phat':phat_bernoulli,'operyr':yy, 'caseid':case_id}))
        holder_metrics.append(pd.DataFrame({'auc':metrics.roc_auc_score(ytest.values, phat_bernoulli),
            'pr':metrics.average_precision_score(ytest.values, phat_bernoulli)},index=[0]))
    holder_vv.append(pd.concat(holder_metrics).assign(outcome=vv,operyr=tmp_train_years))
    holder_phat.append(pd.concat(holder_score).assign(outcome=vv))
    # Save the model
    path_mdl = os.path.join(dir_models,'NB_cpt_' + vv + '.pickle')
    with open(path_mdl, 'wb') as file:
        dill.dump(mdl_bernoulli,file)

res_cpt = pd.concat(holder_vv).reset_index(drop=True)
res_cpt.insert(0,'tt','cpt')

res_phat = pd.concat(holder_phat).reset_index(drop=True)
res_phat.to_csv(os.path.join(dir_output,'nbayes_phat.csv'),index=False)


####################################################
# ---- STEP 3: LEAVE-ONE-YEAR - ALL VARIABLES ---- #

holder_vv = []
for ii, vv in enumerate(cn_Y):
    print('##### ------- Outcome %s (%i of %i) -------- #####' % (vv, ii+1, len(cn_Y)))
    tmp_ii = pd.concat([dat_Y.operyr, dat_Y[vv]==-1],axis=1)
    tmp_ii = tmp_ii.groupby('operyr')[vv].apply(np.sum).reset_index().rename(columns={vv:'n'})
    tmp_years = tmp_ii[tmp_ii.n == 0].operyr.values
    tmp_train_years = tmp_years[tmp_years > tmp_years.min()]
    holder_auc = []
    for yy in tmp_train_years:
        print('Year %i' % (yy))
        idx_train = dat_X.operyr.isin(tmp_years) & (dat_X.operyr < yy)
        idx_test = dat_X.operyr.isin(tmp_years) & (dat_X.operyr == yy)
        Xtrain, Xtest = dat_X.loc[idx_train, cn_X].reset_index(drop=True), \
                        dat_X.loc[idx_test, cn_X].reset_index(drop=True)
        ytrain, ytest = dat_Y.loc[idx_train,vv].reset_index(drop=True), \
                        dat_Y.loc[idx_test,vv].reset_index(drop=True)
        # --- train model --- #
        mdl_bernoulli = mbatch_NB(method='bernoulli')
        mdl_bernoulli.fit(Xtrain,ytrain.values,mbatch=100000)
        phat_bernoulli = mdl_bernoulli.predict(Xtest)[:,1]
        mdl_gaussian = mbatch_NB(method='gaussian')
        holder_auc.append(metrics.roc_auc_score(ytest.values, phat_bernoulli))
    df_ii = pd.DataFrame({'outcome':vv,'operyr':tmp_train_years, 'auc':holder_auc})
    holder_vv.append(df_ii)

res_all = pd.concat(holder_vv).reset_index(drop=True)
res_all.insert(0,'tt','all')

##############################################
# ---- STEP 4: COMBINE RESULTS AND SAVE ---- #

res_both = pd.concat([res_cpt, res_all],axis=0).reset_index(drop=True)
res_both.to_csv(os.path.join(dir_output,'naivebayes_results.csv'),index=False)

# Calculate percentage
res_pct = res_both.pivot_table('auc',['outcome','operyr'],'tt').reset_index()
res_pct['pct'] = 1 - (res_pct.cpt-0.5) / (res_pct['all']-0.5)
res_pct.pct = np.where(res_pct.pct < 0 , 0 ,res_pct.pct)

g = sns.FacetGrid(data=res_both,col='outcome',col_wrap=5,hue='tt')
g.map(sns.scatterplot,'operyr','auc')
g.add_legend()
g.savefig(os.path.join(dir_figures,'auc_naivebayes1.png'))

g = sns.FacetGrid(data=res_pct,col='outcome',col_wrap=5)
g.map(sns.scatterplot,'operyr','pct')
g.savefig(os.path.join(dir_figures,'auc_naivebayes2.png'))