Forecasting the future

[kthouz]

Hi,
First of all thanks for putting this ARIMA summary online.
Second, I have a question. How would someone make predictions on the future (test set) if observed data are still unknown (NaN). Here is a sample code I tried https://github.com/kthouz/Misc/blob/master/arima_tuto.ipynb

[manishajohn]

I have the sales input data for 2 years say for 2014-2015 from month 04(April)-12(Dec) and need to forecast the value for the year 2016.
In this case,i have only small amount of data say less than 50 rows.
For this i am getting the error "ValueError: maxlag should be < nobs" ,so i browsed and found that i need to set the lesser maxlag value,but have no clue where to set it.
Already tried setting it in fit(maxlag=intValue,ic='bic') but it is not getting assigned with the new value and again the same error.
Where i need to and how i should set it actually?
Please help.
Looking for your response.
This is the code i tried:
def test_final_output():
df = pd.read_csv('../../data/salesoneproduct.csv', index_col=0)
nobs=len(df)
df.index.name=None
df.reset_index(inplace=True)
start = datetime.datetime.strptime("2014-04-01", "%Y-%m-%d")
date_list = [start + relativedelta(months=x) for x in range(0,48)]
df['index'] =date_list
df.set_index(['index'], inplace=True)
df.index.name=None
df.columns= ['riders']
df['riders'] = df.riders.apply(lambda x: int(x)*100)
mod = sm.tsa.statespace.SARIMAX(df.riders, trend='n', order=(0,1,0), seasonal_order=(1,1,1,12)
results = mod.fit(ic='bic',maxlag=10)
start = datetime.datetime.strptime("2019-04-01", "%Y-%m-%d")
date_list = [start + relativedelta(months=x) for x in range(0,12)]
future = pd.DataFrame(index=date_list, columns= df.columns)
df = pd.concat([df, future])
df['forecast'] = results.predict(start = 48, end = 59, dynamic= True)
df[['riders', 'forecast']].ix[-24:].plot(figsize=(12, 8))
print (df.forecast/100)

please provide a solution thanks in advance