【发布时间】:2020-06-14 09:58:17
【问题描述】:
我正在训练一个模型来使用 Facebook Prophet 模型预测能耗值。 但是在使用 pickle 保存模型时,我收到了这个错误:
TypeError Traceback (most recent call last)
<ipython-input-54-9ef2990419d8> in <module>
1 with open('forecast_model.pckl', 'wb') as fout:
----> 2 pickle.dump(model, fout)
TypeError: can't pickle _thread.RLock objects
代码:
from fbprophet import Prophet
import pickle
model = Prophet()
model.fit(df)
with open('forecast_model.pkl', 'wb') as fout: <== Error coming here
pickle.dump(model, fout)
df.head()
ds y
0 2018-12-19 15:00:02 250.0
1 2018-12-19 16:00:03 295.0
2 2018-12-19 18:00:02 288.0
3 2018-12-19 19:00:02 286.0
4 2018-12-19 20:00:02 229.0
[编辑] 解决方案链接:https://github.com/facebook/prophet/issues/1361:
下面给出的代码对我有用
model = Prophet()
model.stan_backend.logger = None
with open('test.pickle', 'wb') as f:
pickle.dump(m, f)
【问题讨论】:
标签: python machine-learning pickle facebook-prophet