Closed ayn2 closed 2 years ago
Thanks @ayn2! I presume this error doesn't happen when we run SygnetModel(..., mixed_activation=False)?
Nope, happens still
Not using mixed activation function -- generated data may not conform to real data if it contains categorical columns.
Epoch: 100%|██████████| 1/1 [00:09<00:00, 9.36s/it]
Traceback (most recent call last):
File "C:/Users/Artem/PythonProjects/Public_SyGNet/src/sygnet/test123.py", line 32, in
It does complete the training though, including completing >1 epochs if needed.
No problem, I'll figure it out! Thanks for checking
Hmmm, I'm not able to replicate this one using the code above, and the following imports:
import numpy as np
import pandas as pd
from numpy.random import default_rng
from sygnet import SygnetModel
It may be we've just overlapped on versions. If you're calling this internally, I'm happy to check again with the full dependency list etc. (just reopen this issue)
Epoch: 100%|██████████| 1/1 [00:08<00:00, 8.85s/it] Traceback (most recent call last): File "C:/Users/Artem/PythonProjects/Public_SyGNet/src/sygnet/test123.py", line 32, in
synth_data1 = model.sample(nobs = 1000)
File "C:\Users\Artem\PythonProjects\Public_SyGNet\src\sygnet\sygnet_interface.py", line 309, in sample
n_cat_vars = self.data_encoders[0].n_featuresin
AttributeError: 'OneHotEncoder' object has no attribute 'n_featuresin'
=== Code: rng = default_rng(seed=100) manual_seed(100)
def gen_sim_data(rng, n=100000): x1 = rng.uniform(low=0, high=1, size=n) x2 = rng.uniform(low=0, high=1, size=n) x3 = rng.normal(loc=x1 + x2, scale=0.1) y = rng.normal(loc=3 x1 + 2 x2 + 1, scale=1)
train_data = gen_sim_data(rng) train_data.head()
model = SygnetModel(mode = "wgan") model.fit(data = train_data, epochs = 1) synth_data1 = model.sample(nobs = 1000)
synth_data1.head()