Grid search

Grid search is like a game of find-the-right-thing.

Imagine you are playing a game in your room and your mommy hides your favorite toy somewhere. You want to find your toy but you don't know where it is. So, mommy tells you that she hid it in one of the four corners of the room.

Now, you start looking for your toy by checking all the four corners one-by-one. This is called a grid search. You go to the first corner and check if your toy is there. You don't find it so you go to the next corner and check again. You keep doing this until you find your toy.

In the same way, when we have a problem of finding the best model for our data, we use a grid search. We have different models and each model has its own set of parameters. We want to find the best model and the best set of parameters that give us the highest accuracy.

So we make a grid with all the combinations of parameters for each model. Then, we train and test each model with each set of parameters. Finally, we pick the model and the set of parameters that give us the best accuracy.

Just like you found your favorite toy in one of the four corners, we find the best model and the best set of parameters with a grid search.