0

当我复制这个（下面的源代码）时，我总是看到同样的错误。

我使用了 pycharm IDE，并使用 anaconda 安装了先知（pip 不起作用，但 conda-forge 起作用）

来源：https ://colab.research.google.com/drive/1NN_vY_hp9gmHfqqRi778-V_7PRRXG8ww

import pandas as pd
import plotly.graph_objs as go
import plotly.offline as py
from fbprophet import Prophet
from fbprophet.plot import plot_plotly, add_changepoints_to_plot
py.init_notebook_mode()

import numpy as np

# Confirmation, recovery, and death data sets by region worldwide
url = 'https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv'
data = pd.read_csv(url, error_bad_lines=False)

# Understanding the structure of the data set
data.head()

# Make Korea's confirmed cases timeseries dataframe

df_korea = data[data['Country/Region'] == 'Korea, South']
df_korea = df_korea.T[4:]

df_korea = df_korea.reset_index().rename(columns={'index': 'date', 155: 'confirmed'})

df_korea.tail()

# Plot Korean COVID19 confirmed cases.

fig = go.Figure()

fig.add_trace(
    go.Scatter(
        x=df_korea.date,
        y=df_korea.confirmed,
        name='Confirmed in Korea'
    )
)

fig

# Make dataframe for Facebook Prophet prediction model.
df_prophet = df_korea.rename(columns={
    'date': 'ds',
    'confirmed': 'y'
})

df_prophet.tail()

# Make Prophet model including daily seasonality
m = Prophet(
    changepoint_prior_scale=0.5, # increasing it will make the trend more flexible
    changepoint_range=0.95, # place potential changepoints in the first 98% of the time series
    yearly_seasonality=False,
    weekly_seasonality=True,
    daily_seasonality=True,
    seasonality_mode='additive'
)

m.fit(df_prophet)

future = m.make_future_dataframe(periods=7)
forecast = m.predict(future)

fig = plot_plotly(m, forecast)
py.iplot(fig)

fig = m.plot(forecast)
a = add_changepoints_to_plot(fig.gca(), m, forecast)

错误信息：

Traceback (most recent call last):
  File "C:/Users/daystd/PycharmProjects/covid_forecasting/covid.py", line 117, in <module>
    forecast = m.predict(future)
  File "F:\Anaconda\envs\pyqt_env\lib\site-packages\fbprophet\forecaster.py", line 1204, in predict
    intervals = self.predict_uncertainty(df)
  File "F:\Anaconda\envs\pyqt_env\lib\site-packages\fbprophet\forecaster.py", line 1435, in predict_uncertainty
    sim_values = self.sample_posterior_predictive(df)
  File "F:\Anaconda\envs\pyqt_env\lib\site-packages\fbprophet\forecaster.py", line 1393, in sample_posterior_predictive
    s_m=component_cols['multiplicative_terms'],
  File "F:\Anaconda\envs\pyqt_env\lib\site-packages\fbprophet\forecaster.py", line 1464, in sample_model
    trend = self.sample_predictive_trend(df, iteration)
  File "F:\Anaconda\envs\pyqt_env\lib\site-packages\fbprophet\forecaster.py", line 1501, in sample_predictive_trend
    n_changes = np.random.poisson(S * (T - 1))
  File "mtrand.pyx", line 3592, in numpy.random.mtrand.RandomState.poisson
  File "_common.pyx", line 865, in numpy.random._common.disc
  File "_common.pyx", line 414, in numpy.random._common.check_constraint
ValueError: lam value too large

Process finished with exit code 1

我在谷歌上找不到任何关于这种错误的信息。

我需要帮助 ：（

(ValueError: lam 值太大)

我认为这段代码在以前版本的先知中有效，但现在不行。

所以我想知道如何解决它或通过 conda 安装过去的先知。我是python的初学者。所以我不知道如何通过“conda -c conda-forge fbprophet”的转换来安装以前版本的包

4

1 回答 1

0

您可以在此处找到此错误的描述：https ://github.com/facebook/prophet/issues/1559

'ds' 列应该是 Pandas 期望的日期戳，什么是日期的 YYYY-MM-DD。您可以添加此行：

# Make dataframe for Facebook Prophet prediction model.
df_prophet = df_korea.rename(columns={
    'date': 'ds',
    'confirmed': 'y'})

df_prophet['ds'] = pd.to_datetime(df_prophet['ds'])

这应该可以解决您的问题。

如果无论如何你想安装以前版本的 fbprophet，你不能使用 conda-forge 来做到这一点，因为它总是得到最新版本。

您应该从这里下载历史版本：https ://pypi.org/project/fbprophet/#history并运行conda install package.tar.gz

于 2022-01-30T11:31:12.043 回答