def get_gam_model(self, features: [Field], model_type=TYPE_LINEAR):
model_spec = f(0) if features[0].is_factor() else s(
0, n_splines=self.num_splines)
for i in range(1, len(features)):
model_spec += f(i) if features[i].is_factor() else s(
i, n_splines=self.num_splines)
if model_type == TYPE_LINEAR:
return LinearGAM(model_spec)
if model_type == TYPE_LOGISTIC:
return LogisticGAM(model_spec)
def GAM_linear(X, y):
X= X.to_numpy()
y = y.to_numpy()
from pygam import LinearGAM, s, f, te
gam = LinearGAM(s(0) +s(1) +f(2))
gam.gridsearch(X,y)
y_pred = gam.predict(X)
y_pred = pd.DataFrame(y_pred)
y_pred['actual'] =y
y_pred['residual'] = y_pred.actual-y_pred[0]
return gam, gam.summary(), y_pred
def _fit(self, X, y, mylam=None, **kwargs):
if isinstance(X, pd.DataFrame):
X = X.values
if not self.fit_binary_feat_as_factor_term:
self.model = self.model_cls(max_iter=self.max_iter,
n_splines=self.n_splines,
**self.kwargs)
else:
formulas = []
for idx, feat_name in enumerate(self.feature_names):
num_unique_x = len(self.X_values_counts[feat_name])
if num_unique_x < 2:
continue
if num_unique_x == 2:
formulas.append(f(idx))
else:
formulas.append(s(idx))
the_formula = formulas[0]
for term in formulas[1:]:
the_formula += term
self.model = self.model_cls(the_formula,
max_iter=self.max_iter,
n_splines=self.n_splines,
**self.kwargs)
if not self.search:
# Just fit the model with this lam
return self.model.fit(X, y, **kwargs)
if mylam is None:
mylam = self.search_lam
# do a grid search over here
try:
print('search range from %f to %f' % (mylam[0], mylam[-1]))
self.model.gridsearch(X, y, lam=mylam, **kwargs)
except (np.linalg.LinAlgError, pygam.utils.OptimizationError) as e:
print('Get the following error:', str(e),
'\nRetry the grid search')
if hasattr(self.model, 'coef_'):
del self.model.coef_
self._fit(X, y, mylam=mylam[1:], **kwargs)
if not hasattr(self.model,
'statistics_'): # Does not finish the training
raise Exception('Training fails.')
return self
def fit(self):
S = s(0) if self.feature_names[0] in self.numerical_features else f(0)
for i in range(1, len(self.feature_names)):
if self.feature_names[i] in self.numerical_features:
S += s(i)
else:
S += f(i)
if self.mode == 'regression':
gam = LinearGAM(S)
gam.gridsearch(self.X_train, self.y_train)
self._is_fitted = True
self.explainer = gam
elif self.mode == 'classification':
gam = LogisticGAM(S)
gam.gridsearch(np.array(self.X_train), self.y_train)
self._is_fitted = True
self.explainer = gam
else:
raise NameError(
'ERROR: mode should be regression or classification')
def lingam(term='spline'):
"""
Method to load unfitted Generalized Additive Models models of
type modelclass
INPUT:
term: 'linear', 'spline' or 'factor'
RETURN:
model
"""
if term is 'linear':
regmod = LinearGAM(l(0))
# GAM with spline term
elif term is 'spline':
regmod = LinearGAM(s(0))
# GAM with factor term
elif term is 'factor':
regmod = LinearGAM(f(0))
else:
raise ValueError('Given Gam term unknown')
utils.display_get_params('LinearGAM Model Description',
regmod.get_params())
return(regmod)
gam2.summary()
import pandas as pd
pd.DataFrame(X).corr()
######################################################
# regression
from pygam import LinearGAM, s, f
from pygam.datasets import wage
X, y = wage(return_X_y=True)
## model
gam = LinearGAM(s(0) + s(1) + f(2))
gam.gridsearch(X, y)
gam.summary()
## plotting
plt.figure()
fig, axs = plt.subplots(1, 3)
titles = ['year', 'age', 'education']
for i, ax in enumerate(axs):
XX = gam.generate_X_grid(term=i)
ax.plot(XX[:, i], gam.partial_dependence(term=i, X=XX))
ax.plot(XX[:, i],
gam.partial_dependence(term=i, X=XX, width=.95)[1],
c='r',
ls='--')
from pygam.datasets import wage
from pygam import LinearGAM, s, f
import numpy as np
import matplotlib.pyplot as plt
X, y = wage()
gam = LinearGAM(s(0, n_splines=5) + s(1) + f(2)).fit(X, y)
gam.summary()
lam = np.logspace(-3, 5, 5)
lams = [lam] * 3
gam.gridsearch(X, y, lam=lams)
gam.summary()
lams = np.random.rand(100, 3) # random points on [0, 1], with shape (100, 3)
lams = lams * 8 - 3 # shift values to -3, 3
lams = np.exp(lams) # transforms values to 1e-3, 1e3
random_gam = LinearGAM(s(0) + s(1) + f(2)).gridsearch(X, y, lam=lams)
random_gam.summary()
print(gam.statistics_['GCV'] < random_gam.statistics_['GCV'])
for i, term in enumerate(gam.terms):
if term.isintercept:
continue
XX = gam.generate_X_grid(term=i)
#https://pygam.readthedocs.io/en/latest/notebooks/quick_start.html# import pygam from pygam.datasets import wage X, y = wage() from pygam import LinearGAM, s, f #Let’s fit a spline term to the first 2 features, and a factor term to the 3rd feature. gam = LinearGAM(s(0) + s(1) + f(2)).fit(X, y) gam.summary()
# https://pygam.readthedocs.io/en/latest/notebooks/quick_start.html from pygam.datasets import wage X, y = wage() X.shape, y.shape from pygam import LinearGAM, s, f gam = LinearGAM(s(0) + s(1) + f(2)).fit(X, y) gam.summary() #by deafaule s has 20 base functions gam2 = LinearGAM(s(0, n_splines=5) + s(1) + f(2)).fit(X, y) gam2.summary() #by default all terms has lambda penalty of 0.6, #running grid search for lambda optimization #using GCV (generalized cv-score) import numpy as np lam = np.logspace(-3, 5, 5) lams = [lam] * 3 lams gam3 = LinearGAM(s(0) + s(1) + f(2)) gam3.gridsearch(X, y, lam=lams) gam3.summary()
def gam_wave_1(df, df_train: pd.DataFrame):
galaxy_to_int = dict(
(i, g) for g, i in enumerate(df_train.galaxy.unique()))
int_to_galaxy = {v: k for k, v in galaxy_to_int.items()}
tmp_df_to_predict = df.loc[:, predictors_wave_1].dropna()
tmp_df_to_change_train = df_train.loc[:, predictors_wave_1].dropna()
df_train_return = df_train.copy()
df_train.loc[:, ['galaxy']] = [galaxy_to_int[i] for i in df_train.galaxy]
for TARGET in wave_1_gam:
tmp_df_to_train_on = df_train.loc[:, predictors_wave_1 + [TARGET]]
tmp_df_to_train_on = tmp_df_to_train_on.dropna()
X = tmp_df_to_train_on.loc[:, tmp_df_to_train_on.columns != TARGET]
y = tmp_df_to_train_on.loc[:, TARGET]
lams = np.exp(np.random.random(size=(500, 9)) * 6 - 3)
gam = LinearGAM(
s(0, dtype='categorical', by=1) + f(0) + s(1) + s(2) + s(3) +
s(4) + s(5) + s(6) + s(7)).gridsearch(np.array(X),
np.array(y),
lam=lams)
tmp_df_to_predict.loc[:, ['galaxy']] = [
galaxy_to_int[i] for i in tmp_df_to_predict.galaxy
]
tmp_df_to_change_train.loc[:, ['galaxy']] = [
galaxy_to_int[i] for i in tmp_df_to_change_train.galaxy
]
tmp_df_to_predict[TARGET] = gam.predict(tmp_df_to_predict)
tmp_df_to_change_train[TARGET] = gam.predict(tmp_df_to_change_train)
tmp_df_to_predict.loc[:, ['galaxy']] = [
int_to_galaxy[i] for i in tmp_df_to_predict.galaxy
]
tmp_df_to_change_train.loc[:, ['galaxy']] = [
int_to_galaxy[i] for i in tmp_df_to_change_train.galaxy
]
df = df.set_index(['galaxy', 'galactic year']).fillna(
tmp_df_to_predict.set_index(['galaxy',
'galactic year'])).reset_index()
df_train_return = df_train_return.set_index([
'galaxy', 'galactic year'
]).fillna(tmp_df_to_change_train.set_index(['galaxy', 'galactic year'
])).reset_index()
tmp_df_to_predict = tmp_df_to_predict.drop(columns=[TARGET], axis=1)
tmp_df_to_change_train = tmp_df_to_change_train.drop(columns=[TARGET],
axis=1)
return df, df_train_return
#'inverse'
#'log'
#'inverse-squared'
from pygam.datasets import wage
import matplotlib.pyplot as plt
X, y = wage(return_X_y=True)
#X[0] es el año X[0] = 0 es 2000?...
#X[1] es la edad de la persona
#X[2] es su nivel de estudios, 0 = basica, 1=media superior, 2 = universidad, 3= posgrado
#y ingresos $$
## model
gam1 = LinearGAM(s(0) + s(1) + f(2), fit_intercept=False)
gam1.gridsearch(X, y)
## plotting
plt.figure(figsize=(10, 7.5))
fig, axs = plt.subplots(1, 3)
titles = ['year', 'age', 'education']
for i, ax in enumerate(axs):
XX = gam1.generate_X_grid(term=i)
ax.plot(XX[:, i], gam1.partial_dependence(term=i, X=XX))
ax.plot(XX[:, i],
gam1.partial_dependence(term=i, X=XX, width=.95)[1],
c='r',
ls='--')
ax.set_title(titles[i])
plt.rcParams['figure.figsize'] = [10, 7.5]
df = pd.DataFrame(dataset, columns=column_name)
# df1 = pd.DataFrame(dataset1, columns=column_name)
db.connect.commit()
train_value = df['2020-09-01' > df['date']]
x_train1 = train_value.iloc[:, 1:].astype('float64')
y_train1 = train_value['value'].astype('float64').to_numpy()
x_train2 = train_value['rain'].astype('float64')
from pygam import LinearGAM, s, f
from pygam.datasets import wage
x_train2, y_train1 = wage()
gam = LinearGAM(s(0) + s(1) + f(2)).fit(x_train2,
y_train1) # s(0) + s(1) + f(2)
gam.summary()
import matplotlib.pyplot as plt
for i, term in enumerate(gam.terms):
if term.isintercept:
continue
XX = gam.generate_X_grid(term=i)
pdep, confi = gam.partial_dependence(term=i, X=XX, width=0.95)
plt.figure()
plt.plot(XX[:, term.feature], pdep)
plt.plot(XX[:, term.feature], confi, c='r', ls='--')
import pygam as pg
import pandas as pd
import matplotlib.pyplot as plt
filtered = pd.read_csv('full.csv')
filtered = filtered.query('score != 0.5').query('time_awake<24')
gam = pg.LogisticGAM(
pg.s(0, basis='cp') + pg.f(1) + pg.s(2) + pg.s(3) + pg.s(4) +
pg.s(5)).fit(X=filtered[[
'hour', 'day', 'within_day', 'last_3_days', 'rating_diff', 'time_awake'
]],
y=filtered['win'])
gam.summary()
##hour has to be a cyclical smoother
def draw_terms(gam, plt=None):
for i, term in enumerate(gam.terms):
if term.isintercept:
continue
XX = gam.generate_X_grid(term=i)
pdep, confi = gam.partial_dependence(term=i, X=XX, width=0.95)
plt.figure()
plt.plot(XX[:, term.feature], pdep)
plt.plot(XX[:, term.feature], confi, c='r', ls='--')
plt.title(repr(term))
plt.show()
elif row['maritl'] == '2. Married':
val = 1
elif row['maritl'] == '3. Widowed':
val = 2
elif row['maritl'] == '4. Divorced':
val = 3
else:
val = 4
return val
Wage['mar_d'] = Wage.apply(marital_d, axis=1)
Wage['job_d'] = np.where(Wage['jobclass'] == '1. Industrial', 0, 1)
x3 = pd.concat([Wage['age'], Wage['mar_d'], Wage['job_d']], axis=1)
gam = LinearGAM(s(0, n_splines=4) + f(1) + f(2)).fit(x3, Wage['wage'])
for i, term in enumerate(gam.terms):
if term.isintercept:
continue
XX = gam.generate_X_grid(term=i)
pdep, confi = gam.partial_dependence(term=i, X=XX, width=0.95)
plt.figure()
plt.plot(XX[:, term.feature], pdep)
plt.plot(XX[:, term.feature], confi, c='r', ls='--')
plt.title(repr(term))
plt.show()
# ----------------------------------------------------------------------------