Building Time Series Model - Getting error involving datetime

Question

CODE SNIPPETS:

...

#get predictions starting from 2016-01-01 and calculate confidence intervals
        pred = output.get_prediction(start=pd.to_datetime('2017-09-14'), dynamic=False)
        pred_conf = pred.conf_int()

ERROR:

TypeError: int() argument must be a string, a bytes-like object or a number, not 'Timestamp'

Comments

Hey, @Sharonda,

Could you please how did you try to run your code? And along with can you please also post your models.py?

Answer 1

Hey sharonda,

Use a python datetime or string 

results.get_prediction(start='2020-04-08', axis=0)
# or
import datetime as dt
results.get_prediction(start=dt.datetime(2020, 4, 8), axis=0)

Try this once or post your model.py file.

Thank You!!

Comments

Thank you, Niroj the following statement cleared the error:

---

results.get_prediction(start='2020-04-08', axis=0)

---

Hello I am experiencing a similar error and tried the same thing but did not work. Can you please guide me on this?

#code

pred = results.get_prediction(start=pd.to_datetime('2019-09-25'), dynamic=False, axis = 0)
pred_ci = pred.conf_int()
ax = y[:].plot(label='observed')
pred.predicted_mean.plot(ax=ax, label='One-step ahead Forecast', alpha=.7, figsize=(14, 4))
ax.fill_between(pred_ci.index,
                pred_ci.iloc[:, 0],
                pred_ci.iloc[:, 1], color='k', alpha=.2)
ax.set_xlabel('Date')
ax.set_ylabel('Close')
plt.legend()
plt.show()

ERROR 

TypeError: int() argument must be a string, a bytes-like object or a number, not 'Timestamp'

---

Hello Sanket Dayama,

You have two options:

1. You can call results.predict using integers for start and end (e.g. results.predict(start=results.nobs, end=results.nobs + 10)) and then attach whatever dates you like to the resulting forecasts.

2. You can reindex your data to have a date series with daily frequency. For example:

train1_reindex = train1.reindex(pd.DatetimeIndex(start=train1.date[0], end=train1.date[-1], freq='D'))

This will mean your new time series will have NaNs in it, but that's not a problem for SARIMAX. In fact, it should give you better results, since simply removing missing observations is not the right way to deal with missing observations in models like ARIMA where today's value depends on yesterday's value.

Hope this helps!!