Setting a random state for reproducibility
Before we start discussing the steps and models you may use in a tabular competition, it will be useful to return to the theme of reproducibility we mentioned above.
In most of the commands in the code you see on Kaggle Notebooks, you will find a parameter declaring a number, a seed, as the random state. This setting is important for the reproducibility of your results. Since many algorithms are not deterministic but are based on randomness, by setting a seed you influence the behavior of the random generator, making it predictable in its randomness: the same random seed corresponds to the same sequence of random numbers. In other words, it allows you to obtain the same results after every run of the same code.
That is why you find a random seed setting parameter in all machine learning algorithms in Scikit-learn as well as in all Scikit-learn-compatible models (for instance, XGBoost, LightGBM, and CatBoost, to name the most popular...
