Statsmodels: ValueError: Buffer dtype mismatch, expected 'double' but got 'long'

[AshtonDu]

Issue Description

I'm using models from statsmodels.tsa to do time series analysis and I ran into issue of : ValueError: Buffer dtype mismatch, expected 'double' but got 'long'.

Environment:
MacOS 10.14.5
Anaconda
Syper 3.3.5
Python 3.7.2
statsmodels 0.10.0

Code:

import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error
from math import sqrt
from statsmodels.tsa.api import ExponentialSmoothing, SimpleExpSmoothing, Holt
import matplotlib.pylab as plt
import matplotlib.dates as mdates

df = pd.read_csv('data.csv')

dateparse = lambda dates: pd.datetime.strptime(dates, '%Y-%m-%d')
df = pd.read_csv('data.csv', 
                   parse_dates=['Month'], 
                   index_col='Month',
                   date_parser=dateparse)

df = df.sort_index(axis = 0, ascending=True)

train_rev = df['Revenue'].values[:33]
train_index = pd.date_range('2016-07-01', periods = 33, freq = 'MS')
train = pd.Series(train_rev, train_index)

test_rev = df['Revenue'].values[33:]
test_index = pd.date_range('2019-04-01', periods = 3, freq = 'MS')
test = pd.Series(test_rev, test_index)

ses = SimpleExpSmoothing(train)
ses = ses.fit()

Its giving me this ValueError every time I run the code. To test, I have copied codes from time series analysis tutorials of statsmodels (link: https://www.statsmodels.org/dev/examples/notebooks/generated/exponential_smoothing.html). I can successfully get the final result without any errors. But when I run my code. Its giving me the error.

Would love to know how to fix the issue?
Thanks for your time and help.

[bashtage]

You should explicitly cast your data to double, as in train = train.astype('double)`

[AshtonDu]

Thanks for your reply. This solved the issue!

[bashtage]

Thanks for reporting, fixed in master.