import sys, numpy as np, pandas as pd
from scipy import stats
sys.path.insert(0, 'scripts')
d = pd.read_csv(sys.argv[1])
d['secs'] = d.clock.str.split(':').apply(lambda x: int(x[0])*60+float(x[1]))
d['t'] = (d.quarter-1)*720 + (720-d.secs)
d = d.sort_values(['t']).reset_index(drop=True)
# putback = follow-up attempt by same team within 3 s of that team's previous missed shot
prev_t = d.groupby('team').t.shift(); prev_res = d.groupby('team').result.shift()
d['putback'] = (d.t - prev_t <= 3) & (prev_res == 'miss')
s = d[(d.team=='SAS') & (~d.putback)].copy()
print('SAS putbacks dropped:', int(((d.team=='SAS')&d.putback).sum()), 'remaining', len(s))
y = (s.result=='make').astype(float).values
h = (s.quarter>=3).astype(float).values
mid = ((s.shot_type=='2pt')&(s.distance_ft>3)).astype(float).values
th = (s.shot_type=='3pt').astype(float).values
def fit(X):
    b=np.zeros(X.shape[1])
    for _ in range(100):
        p=1/(1+np.exp(-X@b)); H=X.T@(X*(p*(1-p))[:,None]); b+=np.linalg.solve(H,X.T@(y-p))
    p=1/(1+np.exp(-X@b)); H=X.T@(X*(p*(1-p))[:,None]); se=np.sqrt(np.diag(np.linalg.inv(H)))
    return b,se,np.sum(y*np.log(p)+(1-y)*np.log(1-p))
X=np.column_stack([np.ones(len(y)),h,mid,th]); b,se,ll=fit(X); _,_,ll0=fit(np.column_stack([np.ones(len(y)),mid,th]))
print('2H OR %.3f CI %.3f-%.3f p_wald %.5f p_LR %.5f'%(np.exp(b[1]),np.exp(b[1]-1.96*se[1]),np.exp(b[1]+1.96*se[1]),2*stats.norm.sf(abs(b[1]/se[1])),stats.chi2.sf(2*(ll-ll0),1)))
for hh in [0,1]:
    g=s[(s.quarter>=3)==bool(hh)]; r=g[(g.shot_type=='2pt')&(g.distance_ft<=3)]
    print('half',hh,'FG',int((g.result=='make').sum()),'/',len(g),' rim',int((r.result=='make').sum()),'/',len(r))
