How can I get reproducible results?

[YixinNJU]

Hi, I am using SOM to do clustering analysis. After that, I calculate metrices to evaluate the performance (Silhouette value and so on), I find the performance metrices changed when I rerun SOM. I then pick the matrices belonging to the same SOM node and composite (average) them for both two runs. I find them differ. I am wondering why this would happen? Is there any way to make the results reproducible? Thank you!

[JustGlowing]

hi @YixinNJU, you can set the parameter random_seed to a given integer.

[YixinNJU]

hi @YixinNJU, you can set the parameter random_seed to a given integer.

Thank you! It works!