def build(self):
train, y, test, _ = data.get()
#
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
for c in ['ap_hi', 'ap_lo', 'height', 'weight']:
squish(df, c, c + 'S', 10)
for c in ['gluc', 'cholesterol']:
squish(df, c, c + 'S', 1)
df = df.drop(to_drop, axis=1)
train = df[:ntrain]
test = df[ntrain:].copy()
#
train = pd.concat([train, y], axis=1)
global_avg = y.mean()
for c in test.columns:
d = wa_dict(train, c, 'cardio', global_avg, 10)
test[c] = test[c].map(d).astype('float32')
test.fillna(global_avg, inplace=True)
kf = model_selection.KFold(n_splits=10, shuffle=True, random_state=1)
for itrain, iset in kf.split(train, y):
for c in test.columns:
d = wa_dict(train.ix[itrain], c, 'cardio', global_avg, 10)
train.ix[iset, c] = train.ix[iset, c].map(d)
train.drop('cardio', axis=1, inplace=True)
train.fillna(global_avg, inplace=True)
return train.astype('float32'), test.astype('float32'), None
def build(self):
train, _, test, _ = data.get()
#
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
h = df.height / 100
df['bwi'] = df['weight'] / (h * h)
df['ap_p'] = df['ap_hi'] - df['ap_lo']
df['ap_m4'] = (df['ap_hi'] + 3 * df['ap_lo']) / 4
df['ap_m3'] = (df['ap_hi'] + 2 * df['ap_lo']) / 3
df['ap_m2'] = (df['ap_hi'] + df['ap_lo']) / 2
male = df['gender'] == 2
df.ix[male, 'ap_hi_e'] = 109 + 0.5 * df.ix[
male, 'age'] / 365.25 + 0.1 * df.ix[male, 'weight']
df.ix[male, 'ap_lo_e'] = 74 + 0.1 * df.ix[
male, 'age'] / 365.25 + 0.15 * df.ix[male, 'weight']
df.ix[~male, 'ap_hi_e'] = 102 + 0.7 * df.ix[
~male, 'age'] / 365.25 + 0.15 * df.ix[~male, 'weight']
df.ix[~male, 'ap_lo_e'] = 78 + 0.17 * df.ix[
~male, 'age'] / 365.25 + 0.1 * df.ix[~male, 'weight']
df = df.drop(to_drop, axis=1)
ftrain = df[:ntrain]
ftest = df[ntrain:]
return ftrain, ftest, None
def build(self):
train, _, test, _ = data.get()
#
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
h = df.height / 100
df['bwi'] = df['weight'] / (h * h)
df['ap_p'] = df['ap_hi'] - df['ap_lo']
df['ap_m4'] = (df['ap_hi'] + 3 * df['ap_lo']) / 4
df['ap_m3'] = (df['ap_hi'] + 2 * df['ap_lo']) / 3
df['ap_m2'] = (df['ap_hi'] + df['ap_lo']) / 2
male = df['gender'] == 2
df.ix[male, 'ap_hi_e'] = 109 + 0.5 * df.ix[male, 'age'] / 365.25 + 0.1 * df.ix[male, 'weight']
df.ix[male, 'ap_lo_e'] = 74 + 0.1 * df.ix[male, 'age'] / 365.25 + 0.15 * df.ix[male, 'weight']
df.ix[~male, 'ap_hi_e'] = 102 + 0.7 * df.ix[~male, 'age'] / 365.25 + 0.15 * df.ix[~male, 'weight']
df.ix[~male, 'ap_lo_e'] = 78 + 0.17 * df.ix[~male, 'age'] / 365.25 + 0.1 * df.ix[~male, 'weight']
df.ix[male, 'weight_ah'] = (df.ix[male, 'ap_hi'] - 109 - 0.5 * df.ix[male, 'age'] / 365.25) / 0.1
df.ix[male, 'weight_al'] = (df.ix[male, 'ap_lo'] - 74 - 0.1 * df.ix[male, 'age'] / 365.25) / 0.15
df.ix[~male, 'weight_ah'] = (df.ix[~male, 'ap_hi'] - 102 - 0.7 * df.ix[~male, 'age'] / 365.25) / 0.15
df.ix[~male, 'weight_al'] = (df.ix[~male, 'ap_lo'] - 78 - 0.17 * df.ix[~male, 'age'] / 365.25) / 0.1
df['ap_hi_ed'] = df['ap_hi'] - df['ap_hi_e']
df['ap_lo_ed'] = df['ap_lo'] - df['ap_lo_e']
df['dw_ah'] = df['weight'] - df['weight_ah']
df['dw_al'] = df['weight'] - df['weight_al']
df = df.drop(to_drop, axis=1)
ftrain = df[:ntrain]
ftest = df[ntrain:]
return ftrain, ftest, None
def predict():
saved = state.load('model')
#saved = None
if saved == None:
train, y, test, _ = data.get()
z = pd.DataFrame()
z['id'] = test.id
z['y'] = 0
v = pd.DataFrame()
v['id'] = train.id
v['y'] = y
cv, _ = run(train, y, test, v, z)
state.save('model', (v, z, cv, None))
else:
v, z, cv, _ = saved
return v, z, cv, _
def predict():
saved = state.load('model')
#saved = None
if saved == None:
train, y, test, _ = data.get()
z = pd.DataFrame()
z['id'] = test.id
z['y'] = 0
v = pd.DataFrame()
v['id'] = train.id
v['y'] = y
cv, _ = run(train, y, test, v, z)
state.save('model', (v, z, cv, None))
else:
v, z, cv, _ = saved
return v, z, cv, _
def build(self):
train, _, test, _ = data.get()
cset = []
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
for sc in ['height', 'weight', 'ap_hi', 'ap_lo']:
tc = df[sc].apply(str)
maxc = tc.apply(len).max()
for n in range(maxc):
df['ft_l_'+sc+'_'+str(n)] = tc.apply(lambda s:ord(s[n]) if n < len(s) else -1)
df['ft_r_'+sc+'_'+str(n)] = tc.apply(lambda s:ord(s[-n]) if n < len(s) else -1)
cset.append('ft_l_'+sc+'_'+str(n))
cset.append('ft_r_'+sc+'_'+str(n))
df = df.drop(to_drop, axis=1)
self.train_= df[:ntrain]
self.test_ = df[ntrain:]
return self.train_, self.test_, None
def build(self):
train, y, test, _ = data.get()
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
dcn = []
for n in [2, 5, 10, 15, 25]:
cname = 'kmeans_' + str(n)
dcn.append(cname)
df[cname] = cluster.KMeans(n_clusters=n).fit_predict(df)
df = pd.get_dummies(df, columns=dcn)
df = df.drop(to_drop, axis=1)
train = df[:ntrain]
test = df[ntrain:].copy()
return train.astype('int32'), test.astype('int32'), None
def build(self):
train, _, test, _ = data.get()
cset = []
ntrain = len(train)
df = pd.concat([train, test], axis=0)
to_drop = df.columns
for sc in ['height', 'weight', 'ap_hi', 'ap_lo']:
tc = df[sc].apply(str)
maxc = tc.apply(len).max()
for n in range(maxc):
df['ft_l_' + sc + '_' + str(n)] = tc.apply(
lambda s: ord(s[n]) if n < len(s) else -1)
df['ft_r_' + sc + '_' + str(n)] = tc.apply(
lambda s: ord(s[-n]) if n < len(s) else -1)
cset.append('ft_l_' + sc + '_' + str(n))
cset.append('ft_r_' + sc + '_' + str(n))
df = pd.get_dummies(df, columns=cset).drop(to_drop, axis=1)
self.train_ = df[:ntrain]
self.test_ = df[ntrain:]
return self.train_, self.test_, None
def build(self):
train, y, test, _ = data_src.get()
xgb_params = dict(
max_depth=5,
learning_rate=0.005,
subsample=0.7,
gamma=5,
alpha=0.01,
#colsample_bytree = 0.8,
objective='binary:logistic',
eval_metric='logloss',
seed=1,
silent=1)
idx = (test.smoke > 0).values * (test.smoke < 1).values
print('values to restore:', np.sum(idx))
xtrain = pd.concat([train, test[~idx]])
ytrain = xtrain['smoke']
xtrain.drop('smoke', axis=1, inplace=True)
print(xtrain.shape, ytrain.shape, test[idx].shape)
dtrain = xgb.DMatrix(xtrain.values, ytrain.values)
dpred = xgb.DMatrix(test[idx].drop('smoke', axis=1).values)
cv = xgb.cv(params=xgb_params,
dtrain=dtrain,
num_boost_round=10000,
early_stopping_rounds=50,
nfold=10,
seed=1,
metrics='error',
stratified=True)
print('smoke num_boost_rounds =', len(cv))
bst = xgb.train(params=xgb_params,
dtrain=dtrain,
num_boost_round=len(cv))
test.ix[idx, 'smoke'] = bst.predict(dpred)
test['smoke'] = (test['smoke'] > 0.5) * 1
return train, y, test, None
def build(self):
train, y, test, _ = data_src.get()
xgb_params = dict(
max_depth = 5,
learning_rate = 0.005,
subsample = 0.7,
gamma = 5,
alpha = 0.01,
#colsample_bytree = 0.8,
objective = 'binary:logistic',
eval_metric = 'logloss',
seed = 1,
silent = 1
)
idx = (test.smoke > 0).values * (test.smoke < 1).values
print('values to restore:', np.sum(idx))
xtrain = pd.concat([train, test[~idx]])
ytrain = xtrain['smoke']
xtrain.drop('smoke', axis=1, inplace=True)
print(xtrain.shape, ytrain.shape, test[idx].shape)
dtrain = xgb.DMatrix(xtrain.values, ytrain.values)
dpred = xgb.DMatrix(test[idx].drop('smoke', axis=1).values)
cv = xgb.cv(params=xgb_params,
dtrain=dtrain,
num_boost_round=10000,
early_stopping_rounds=50,
nfold=10,
seed=1,
metrics='error',
stratified=True)
print('smoke num_boost_rounds =', len(cv))
bst = xgb.train(params=xgb_params, dtrain=dtrain, num_boost_round=len(cv))
test.ix[idx, 'smoke'] = bst.predict(dpred)
test['smoke'] = (test['smoke'] > 0.5) * 1
return train, y, test, None