def output_func(x):
return 1 / 5 * x**2
def complex_noise(x):
return npr.normal(0, 1, len(x)) * np.abs(np.abs(x + 2) - 2) / 1.5
def complex_noise2(x):
return npr.normal(0, 1, len(x)) * 1 / (np.maximum(np.abs(x), 0.3))
def complex_noise3(x):
return npr.normal(0, 1, len(x)) * (np.sin((x - 5)) * 0.8)
n = 500
trX, trY, teX, teY = generate_data_function(output_func,
complex_noise3,
n,
rate=0.1,
loc=[-3, -2],
yloc=[50, 30])
x, y = torch.Tensor(trX), torch.Tensor(trY)
gaussian = Gaussian()
laplace = Laplace()
adaptive = Adaptive()
# # 2-D linear regression
lr2 = PolyRegression(2)
fit2 = train_regular(lr2,
x,
y,
gaussian,
epoch=1000,
noise = [
indep_noise, linear_noise, exp_noise, unimodal_noise, bimodal_noise,
trimodal_noise
]
noise = [unimodal_noise]
Y_name = ['Poly', 'Sin']
output = [polynomial_outcome, sinusoidal_outcome]
for i in range(len(Y_name)):
for j in range(len(n_name)):
dfs = []
dfs_o = []
for r in range(5):
trX, trY, teX, teY = generate_data_function(output[i],
noise[j],
n,
rate=0.1,
loc=[-2],
yloc=[10])
if i == 0:
degree = 2
else:
degree = 5
df = run_experiment(trX, trY, teX, teY, degree)
df['rep'] = r
dfs_o.append(df)
trX, trY, teX, teY = generate_data_function(output[i],
noise[j],
n,
rate=0.0,
loc=[-2],
return pd.DataFrame(stats)
if __name__ == '__main__':
output_func = polynomial_outcome
noise_func = indep_noise
n_name = ['Indep', 'Linear', 'Exp', 'Uni', 'Bi', 'Tri']
noise = [indep_noise, linear_noise, exp_noise, unimodal_noise, bimodal_noise, trimodal_noise]
Y_name = ['Poly']
output = [polynomial_outcome]
for d in [1,2,3,4,5,6,7,8,9,10]:
for i in range(len(Y_name)):
for j in range(len(n_name)):
df_b = []
for r in range(5):
trX, trY, _, _ = generate_data_function(output[i], noise[j], 1000, rate=0.1, loc=[-2], yloc=[10])
teX, teY, _, _ = generate_data_function(output[i], noise[j], 200, rate=0.1, loc=[-2], yloc=[10])
df = run_experiment(trX, trY, teX, teY, degree=d)
df['rep'] = r
df_b.append(df)
print(Y_name[i] + n_name[j] + 'base' + str(d))
pd.concat(df_b).to_csv('results/6_4/' + Y_name[i] + n_name[j] + 'base' + str(d) + 'Outlier.csv')
for d in [1,2,3,4,5,6,7,8,9,10]:
for i in range(len(Y_name)):
for j in range(len(n_name)):
df_n = []
for r in range(5):
trX, trY, _, _ = generate_data_function(output[i], noise[j], 1000, rate=0.1, loc=[-2], yloc=[10])
teX, teY, _, _ = generate_data_function(output[i], noise[j], 200, rate=0.1, loc=[-2], yloc=[10])