choice-metrics.com

Hello,

When I run the explicit partial profiles, according to the related syntax provided in the manual, I get the following error: The 'candidates' parameter of the Modified Fedorov algorithm is neither specified as a valid number, nor a valid file name or path.

The file is CSV (; delimited), the data format is text, and the structure is the same as the one provided in the manual.

Maybe, providing you with a snippet of my data would help better:

1;0;0;160;2;0;1;17.00-20.00;20;1;1;0;160;2;0;1;17.00-20.00;20
1;0;0;80;1;1;0;14.00-17.00;20;1;1;0;80;1;1;0;14.00-17.00;20
1;0;2/3;40;2;1;0;14.00-17.00;40;1;1;2/3;40;2;1;0;14.00-17.00;40
1;0;1/3;40;2;1;0;11.00-14.00;20;1;1;1/3;40;2;1;0;11.00-14.00;20
1;0;1/3;120;0;0;2;08.00-11.00;60;1;1;1/3;120;0;0;2;08.00-11.00;60
1;0;1;40;1;0;2;14.00-17.00;40;1;1;1;40;1;0;2;14.00-17.00;40
1;0;1;160;2;0;1;11.00-14.00;20;1;1;1;160;2;0;1;11.00-14.00;20
1;0;0;120;1;0;1;17.00-20.00;60;1;1;0;120;1;0;1;17.00-20.00;60
1;0;0;40;0;0;1;11.00-14.00;40;1;1;0;40;0;0;1;11.00-14.00;40

I thank you very much for your time and response!

Beyza

The first column is correct, namely all ones. The second column, however, needs to be the choice task number, so 1, 2, 3, etc. After that, the attributes come.

Further, I am not sure what "17.00-20.00" means, you should use the exact same levels as in the Ngene script, which means you can only use numbers. So also "2/3" cannot be recognised.

Also, make sure that your CSV has a header row, e.g. Resp, Choicetask, alt1.x1, etc. It does not matter which labels you use, they are just there for your convenience.

Finally, I am not sure if semicolon delimited will work but you can try. CSV stands for comma-separated variable and comma delimitation should definitely work.

The error message may relate to the fact that it cannot find the file. Please first open the CSV file in Ngene, and then locally refer to it in your script, e.g. ;alg = mfederov(candidates = filename.csv)

Michiel

Hello,

Thank you very much! May I kindly ask you to elaborate on whether having a run for approximately 5 hours (still continuing) for a dataset of 1000 rows is expected?

In the trace section, what I see is the following:

Warning: One or more attributes will not have level balance with the number of rows specified:
optiona.avg_vehicle_traffic, optiona.insurance_coverage, optiona.working_hours, optiona.compensation_per_km,
optionb.avg_vehicle_traffic, optionb.insurance_coverage, optionb.working_hours, optionb.compensation_per_km

With current evaluation: 133 456.

Thank you very much!

Beyza

Yes the modified Federov algorithm may take some time. Note that you will always need to stop the algorithm manually, so whenever Ngene cannot find a much better design you can stop it. Or you can simply wait until it has evaluated X designs (e.g. 100,000, or 500,000, or 1 mln).

The warning appears because your number of rows is not divisible by the number of levels of the mentioned attributes. This is not a major issue. Please check for attribute level balance in your design and possibly impose constraints if needed, especially for numerical attributes, as the mfederov algorithm will mostly use the outer levels for numerical attributes. You can impose constraints such as working_hours[1,2,3](3-6,3-6,3-6) which ensures that each level appears between 3 and 6 times across all rows in the design.

Michiel