In [106]:

    

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as pyplot
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline


    

In [21]:

    

figure = pyplot.figure()
axe = figure.add_subplot(111, projection='3d')
 
x = np.linspace(-5.12, 5.12, 100) # unterteilt den Bereich in 100 Schnitte, ähnlich: np.arange(-5.12, 5.12, 0.1)
y = np.linspace(-5.12, 5.12, 100)
x, y = np.meshgrid(x, y) # erzeugt ein Koordinatensystem
 
# Nun ohne Schleifen: Wir wenden die NumPy-Funktionen (np.cos statt math.cos und np.pi statt math.pi) 
# auf die NumPy-Arrays an (x und y) und erhalten ein NumPy-Array z zurück
 
# Plotte die drei Variablen (x, y, z) im dreidimensionalen Raum
axe.plot_surface(x, y, z, rstride=1, cstride=1, cmap="jet", linewidth=0, antialiased=False)
 
pyplot.title('Rastrigin-Map')
pyplot.grid(True)
axes = pyplot.gca()
axes.set_xlim([-5.12,5.12])
axes.set_ylim([-5.12,5.12])


    

    
Out[21]:





(-5.12, 5.12)






    

In [125]:

    

from hyperopt import fmin, tpe, hp, STATUS_OK, Trials

fspace = {
    'x': hp.uniform('x', -5, 5),
    'y': hp.uniform('y', -5, 5)
}

def rastiridin(x,y):
    return 20 + x**2 - 10 * np.cos(2 * np.pi * x) + y**2 - 10 * np.cos(2* np.pi * y)

def f(params):
    x = params['x']
    y = params['y']
    val = rastiridin(x,y)
    return {'loss': val, 'status': STATUS_OK}

trials = Trials()
N=5000


    

In [126]:

    

best = fmin(fn=f, 
            space=fspace,
            algo=tpe.suggest,
            max_evals=N,
            trials=trials)

print( 'best:', best)

print ('trials:')
for trial in trials.trials[:2]:
    print(np.min(trials.losses()))


    

    




best: {'x': 0.008697138379726133, 'y': -0.0025483920360279896}
trials:
0.01629109602911072
0.01629109602911072

In [127]:

    

from hyperopt.plotting import main_plot_vars
from hyperopt import base

# setting up your search_space, objective_function, and trials goes here
# see the tutorials to know how

domain = base.Domain(f, fspace)
main_plot_vars(trials, bandit=domain)


    

    




finite loss range 0.01629109602911072 80.38888401494933 -0.9837089039708893






    

In [128]:

    

plt.plot(trials.losses(),'.')


    

    
Out[128]:





[<matplotlib.lines.Line2D at 0x7f54da08f6a0>]






    

In [129]:

    

plt.plot(trials.vals["x"],trials.losses(),'.')


    

    
Out[129]:





[<matplotlib.lines.Line2D at 0x7f54da27bbe0>]






    

In [130]:

    

cm = plt.cm.get_cmap('rainbow')
sc = plt.scatter(trials.vals["x"], trials.vals["y"], c=trials.losses(), cmap=cm)
plt.colorbar(sc)


    

    
Out[130]:





<matplotlib.colorbar.Colorbar at 0x7f54d9ed2a58>






    

In [131]:

    

x,y=np.random.uniform(-5,5,N),np.random.uniform(-5,5,N)
losses=rastiridin(x,y)


    

In [132]:

    

cm = plt.cm.get_cmap('rainbow')
sc = plt.scatter(x, y, c=losses, cmap=cm)
plt.colorbar(sc)


    

    
Out[132]:





<matplotlib.colorbar.Colorbar at 0x7f54d9956f60>






    

In [123]:

    

np.min(losses)


    

    
Out[123]:





4.031472071114029

In [158]:

    

from sklearn import svm, datasets
from sklearn.model_selection import GridSearchCV
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import RandomForestRegressor

from sklearn.model_selection import train_test_split
iris = datasets.load_iris()
iris = datasets.load_diabetes()


    

In [159]:

    

X_trn,X_val,y_trn,y_val= train_test_split (iris.data,iris.target,shuffle=True)


    

In [160]:

    

def acc_model(params):
    clf = RandomForestRegressor(**params)
    clf.fit(X_trn,y_trn)
    score=clf.score(X_val,y_val)
    return score

param_space = {
    'max_depth': hp.choice('max_depth', range(1,20)),
    'max_features': hp.choice('max_features', ['log2','auto',1.0]),
    'n_estimators': hp.choice('n_estimators', range(10,100)),
    'criterion': hp.choice('criterion', ["mae", "mse"]),
}

best = 0
def f(params):
    global best
    acc = acc_model(params)
    if acc > best:
        best = acc
    print ('new best:', best, params)
    return {'loss': -acc, 'status': STATUS_OK}

trials = Trials()
best_params = fmin(f, param_space, algo=tpe.suggest, max_evals=100, trials=trials)
print ('best:')
print (best_params)


    

    




new best: 0.5139218820954337 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'log2', 'n_estimators': 54}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 3, 'max_features': 'auto', 'n_estimators': 81}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 11, 'max_features': 'auto', 'n_estimators': 32}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 10, 'max_features': 1.0, 'n_estimators': 65}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 10, 'max_features': 'auto', 'n_estimators': 70}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 4, 'max_features': 'log2', 'n_estimators': 56}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 13, 'max_features': 'auto', 'n_estimators': 71}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 11, 'max_features': 'auto', 'n_estimators': 88}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 1, 'max_features': 'auto', 'n_estimators': 62}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 2, 'max_features': 1.0, 'n_estimators': 21}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 3, 'max_features': 'log2', 'n_estimators': 98}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 18, 'max_features': 1.0, 'n_estimators': 59}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 16, 'max_features': 1.0, 'n_estimators': 30}
new best: 0.5556366776419978 {'criterion': 'mse', 'max_depth': 7, 'max_features': 1.0, 'n_estimators': 74}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 1, 'max_features': 'auto', 'n_estimators': 44}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 18, 'max_features': 'log2', 'n_estimators': 38}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 7, 'max_features': 1.0, 'n_estimators': 62}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 12, 'max_features': 1.0, 'n_estimators': 40}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 3, 'max_features': 'auto', 'n_estimators': 33}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 17, 'max_features': 'log2', 'n_estimators': 16}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 14, 'max_features': 1.0, 'n_estimators': 81}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 19, 'max_features': 'auto', 'n_estimators': 81}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 7, 'max_features': 1.0, 'n_estimators': 89}
new best: 0.5556366776419978 {'criterion': 'mae', 'max_depth': 5, 'max_features': 1.0, 'n_estimators': 89}
new best: 0.5561502639317514 {'criterion': 'mae', 'max_depth': 5, 'max_features': 'auto', 'n_estimators': 89}
new best: 0.5561502639317514 {'criterion': 'mae', 'max_depth': 5, 'max_features': 'auto', 'n_estimators': 58}
new best: 0.5561502639317514 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'auto', 'n_estimators': 45}
new best: 0.5561502639317514 {'criterion': 'mae', 'max_depth': 15, 'max_features': 'auto', 'n_estimators': 94}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'auto', 'n_estimators': 73}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'auto', 'n_estimators': 73}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 8, 'max_features': 'auto', 'n_estimators': 49}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'auto', 'n_estimators': 19}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 5, 'max_features': 'auto', 'n_estimators': 57}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'log2', 'n_estimators': 79}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 2, 'max_features': 'auto', 'n_estimators': 36}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 15, 'max_features': 'auto', 'n_estimators': 69}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 4, 'max_features': 'auto', 'n_estimators': 68}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 13, 'max_features': 'log2', 'n_estimators': 47}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 10, 'max_features': 'auto', 'n_estimators': 26}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 11, 'max_features': 'auto', 'n_estimators': 64}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 8, 'max_features': 'auto', 'n_estimators': 86}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 16, 'max_features': 'log2', 'n_estimators': 73}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 5, 'max_features': 'auto', 'n_estimators': 82}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 12, 'max_features': 'auto', 'n_estimators': 35}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 14, 'max_features': 'auto', 'n_estimators': 43}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'log2', 'n_estimators': 53}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 17, 'max_features': 'auto', 'n_estimators': 80}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 19, 'max_features': 'auto', 'n_estimators': 95}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 9, 'max_features': 'log2', 'n_estimators': 97}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 1, 'max_features': 'auto', 'n_estimators': 61}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 4, 'max_features': 1.0, 'n_estimators': 12}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 18, 'max_features': 'auto', 'n_estimators': 93}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 3, 'max_features': 'log2', 'n_estimators': 48}
new best: 0.5573957046805609 {'criterion': 'mse', 'max_depth': 13, 'max_features': 1.0, 'n_estimators': 51}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 10, 'max_features': 'auto', 'n_estimators': 67}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 10, 'max_features': 'auto', 'n_estimators': 67}
new best: 0.5573957046805609 {'criterion': 'mae', 'max_depth': 10, 'max_features': 1.0, 'n_estimators': 27}
new best: 0.5602717663701917 {'criterion': 'mae', 'max_depth': 10, 'max_features': 'auto', 'n_estimators': 23}
new best: 0.5602717663701917 {'criterion': 'mse', 'max_depth': 2, 'max_features': 'log2', 'n_estimators': 10}
new best: 0.5602717663701917 {'criterion': 'mae', 'max_depth': 11, 'max_features': 'auto', 'n_estimators': 28}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 9, 'max_features': 1.0, 'n_estimators': 29}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 12, 'max_features': 1.0, 'n_estimators': 23}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 7, 'max_features': 1.0, 'n_estimators': 99}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 14, 'max_features': 1.0, 'n_estimators': 31}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 16, 'max_features': 1.0, 'n_estimators': 29}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 9, 'max_features': 1.0, 'n_estimators': 50}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 9, 'max_features': 1.0, 'n_estimators': 23}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 9, 'max_features': 1.0, 'n_estimators': 29}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 9, 'max_features': 1.0, 'n_estimators': 24}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 17, 'max_features': 1.0, 'n_estimators': 75}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 15, 'max_features': 'auto', 'n_estimators': 22}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 19, 'max_features': 'auto', 'n_estimators': 60}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 10, 'max_features': 1.0, 'n_estimators': 55}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 8, 'max_features': 'auto', 'n_estimators': 17}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'auto', 'n_estimators': 96}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 18, 'max_features': 'log2', 'n_estimators': 96}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'auto', 'n_estimators': 96}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 1.0, 'n_estimators': 37}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'auto', 'n_estimators': 72}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 3, 'max_features': 'auto', 'n_estimators': 87}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 1, 'max_features': 1.0, 'n_estimators': 91}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'log2', 'n_estimators': 14}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 4, 'max_features': 'auto', 'n_estimators': 15}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 13, 'max_features': 'auto', 'n_estimators': 39}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 10, 'max_features': 1.0, 'n_estimators': 77}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 2, 'max_features': 'auto', 'n_estimators': 52}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 11, 'max_features': 'log2', 'n_estimators': 66}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 7, 'max_features': 'auto', 'n_estimators': 25}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 16, 'max_features': 'auto', 'n_estimators': 85}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 12, 'max_features': 1.0, 'n_estimators': 46}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 15, 'max_features': 'auto', 'n_estimators': 41}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 17, 'max_features': 'log2', 'n_estimators': 34}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 14, 'max_features': 1.0, 'n_estimators': 90}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 6, 'max_features': 'auto', 'n_estimators': 84}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 19, 'max_features': 'auto', 'n_estimators': 11}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 8, 'max_features': 1.0, 'n_estimators': 63}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 1, 'max_features': 'auto', 'n_estimators': 42}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 10, 'max_features': 'log2', 'n_estimators': 20}
new best: 0.5684749001190446 {'criterion': 'mse', 'max_depth': 18, 'max_features': 1.0, 'n_estimators': 18}
new best: 0.5684749001190446 {'criterion': 'mae', 'max_depth': 5, 'max_features': 'auto', 'n_estimators': 98}
best:
{'criterion': 0, 'max_depth': 8, 'max_features': 2, 'n_estimators': 19}

In [ ]:

    




    

In [161]:

    

cm = plt.cm.get_cmap('rainbow')
sc=plt.scatter(trials.vals["n_estimators"],trials.vals["max_depth"],c=trials.losses(),cmap=cm)
plt.colorbar(sc)


    

    
Out[161]:





<matplotlib.colorbar.Colorbar at 0x7f54d9eb1390>






    

In [162]:

    

plt.plot(trials.vals["n_estimators"],trials.losses(),'.')


    

    
Out[162]:





[<matplotlib.lines.Line2D at 0x7f54daae5240>]






    

In [ ]: