The same statistical or machine learning method can be programmed (implemented) in different ways, and this can have an impact on the results. (I'm not referring to programing errors.)

Moreover, the initial start seed can strongly affect a routine - change the start seed and the results may vary substantially.

So, the same method programmed the same way may give different results on the same data if you change the start seed.

Most (hopefully all) statisticians are aware of this, but I suspect most users (e.g., decision makers) are not. "AI" is not immune to this.

✴️ @AI_Python_EN