Kernel Fails when fitting model with FBProphet

Problems and bugs
1

Kernel: Basic Machine, IPython kernel
fbprophet=0.7.1

Anytime I try to fit a model with FBProphet, such as

# fit model
model = Prophet()
model.fit(peyton)

It shows the info

INFO:fbprophet:Disabling daily seasonality. Run prophet with daily_seasonality=True to override this.

And the Kernel fails in less than 10s.

Already opened an issue in Prophet’s GitHub as well.

2

Hi!

  • I’ve installed created a new notebook with Basic machine,
  • installed fbprophet==0.7.1 (pip) with Library Manager,
  • Written the following code:
import pandas as pd
from fbprophet import Prophet

!wget https://raw.githubusercontent.com/PinkWink/DataScience/master/data/07.%20example_wp_peyton_manning.csv
df = pd.read_csv('07. example_wp_peyton_manning.csv')

model = Prophet()
model.fit(df)

future = model.make_future_dataframe(periods=365)
model.predict(future)

It has shown me the same warning but worked and produced a result. Maybe there is some difference that makes the kernel fail in your case?

3

I have tried importing the data in different ways

data_path = 'https://raw.githubusercontent.com/PinkWink/DataScience/master/data/07.%20example_wp_peyton_manning.csv'

peyton = pd.read_csv(data_path)

And

!wget https://raw.githubusercontent.com/PinkWink/DataScience/master/data/07.%20example_wp_peyton_manning.csv

peyton = pd.read_csv('07. example_wp_peyton_manning.csv')

But the error persists.

In order to see if it was due to the number of libraries imported (as these notebook already has quite a few), I have tried importing only Pandas as FBProphet, but the error persists.

Are you using IPython kernel or Datalore kernel? (If IPython kernel, which is the same as me, it may be due to the the fact that my environment, apart from fbprophet=0.7.1 has also installed pmdarima=1.7.1.)

As I was working in Home, I will create a Workspace and create the notebook there to see if the problem lies here.

4

I was using IPython kernel. In Datalore kernel it works as well.

I’ve added pmdarima=1.7.1, it still works: https://view.datalore.jetbrains.com/notebook/3hwBgnR7bexVg9BPMKGotv

1 Like
6

Just added the Notebooks to a Workspace and changed the way the data was being read solved

from fbprophet import Prophet

!wget = 'https://raw.githubusercontent.com/PinkWink/DataScience/master/data/07.%20example_wp_peyton_manning.csv'

data = '07. example_wp_peyton_manning.csv'
peyton = pd.read_csv(data)

And then fitting the model

m = Prophet()
m.fit(peyton)

And the Kernel didn’t fail.