Forecasting using Python(Pytools) in QlikSense Desktop

[manish1]

Hi Team,

I am using QlikSense Desktop latest version and trying to run forecasting using Pytools.Prophet. 

But it's not showing me the output.  Here is the data and function I have used.

PyTools.Prophet([Data_col.autoCalendar.Date],Avg(Shipments),'freq=M'). I am not getting the forecast line.

Here is the data that I am using:

Quarter	Shipments
1/1/1986	4009
2/1/1986	4123
3/1/1986	4493
4/1/1986	4595
5/1/1986	4245
6/1/1986	4321
7/1/1986	4522
8/1/1986	4806
9/1/1986	4799
10/1/1986	4900
11/1/1986	400
12/1/1986	5000
1/1/1987	3000
2/1/1987	2456
3/1/1987	1234
4/1/1987	3456
5/1/1987	7878

It's a monthly data.

Please help.

[y_grynechko]

Hey, did you start: Qlik-Py-Start.bat? 

If you did, do you get any results in the window that should open with start bat. 

Something like this:

[manish1]

Hi, The  Qlik py tools is in C:\Users\user-name\Documents\qlik-py-tools-6.2  folder and I start the file "Qlik-Py-Start.bat".

I start the .bat file and then open the Qliksense Desktop app which is in  C:\Users\user-name\Documents\Qlik\Sense\Apps.

Its just showing this:

2020-01-13 21:49:45,114 - INFO - Adding to capabilities: Prophet_Multivariate(['a_date', 'b_value', 'c_holidays', 'd_added_regressors', 'e_regressor_args', 'f_other_args'])
2020-01-13 21:49:45,114 - INFO - Adding to capabilities: Prophet_Seasonality_Multivariate(['a_season', 'b_time_series', 'c_holidays', 'd_added_regressors', 'e_regressor_args', 'f_other_args'])
2020-01-13 21:49:45,114 - INFO - Adding to capabilities: sklearn_Calculate_Metrics_Sequence(['a_model_name', 'b_key', 'n_features'])
2020-01-13 21:54:15,378 - INFO - ipv6:[::1]:55612 - Capability 'Prophet' called by user Personal\Me from app C:\Users\user-name\Documents\Qlik\Sense\Apps\forecastv2.qvf
2020-01-13 21:54:15,379 - INFO - ExecuteFunction (functionId: 5, _prophet)

[manish1]

Any help?

[outboardpluto]

Hi Manish,

You need to have future dates in the Quarter columns already present for which you want to forecast the data. The corresponding data to these dates would be left as blank or NULL.

If these dummy dates are present in your data then Prophet will be able to fill the forecasted values for those dates.