Hi all,
I've successfully used fbprophet for a quick forecast on a test dataset I called "ret", which was in the following form with dates labeled 'ds' and values 'y':
288 2016-01-01 12.901537 400928
289 2016-02-01 12.932543 413554
290 2016-03-01 13.039184 460093
291 2016-04-01 13.019078 450935
292 2016-05-01 13.063507 471421
I used the following code:
[inline]model=Prophet()
model.fit(ret)
future_data = model.make_future_dataframe(periods=6, freq = 'm')[/inline]
Now I'm trying to do the exact same thing with daily currency exchange data, but Anaconda is stalling out. I'm using Jupyter Notebook, and notice that in the command prompt window the message "initial log joint probability" appears, after which a series of log probabilities are appearing, but nothing ever finishes and I keep having to close out my Jupyter Notebook session. The new data is in the following form, again with columns 'ds' and 'y':
11752 2017.12.11 1.17668
11753 2017.12.12 1.17387
11754 2017.12.13 1.18232
11755 2017.12.14 1.17751
11756 2017.12.15 1.17481
My non-working code is:
[inline]model=Prophet(daily_seasonality=True)
model.fit(df1)
future = model.make_future_dataframe(periods=30, freq='d')[/inline]
Any ideas are much appreciated!