Don't understand the meaning of ''training a dataset''

[luchins]

Hi, I have a set of data. Those data are based on data mining from my website. I have the number of users per month who go to my website ( X ) , and the time they spent on the website on each webpage ( y ) .
Now with this dataset as an example, could you explain to me please what's the meaning of training dataset? If I would ''train'' those data, what should I do? And what is the meaning of the word? A simple example I am noob starting out from somewhere, please. (I know somenthing about linear regression and logistic regression, anyway never dealt with machine learning, if you could make an example with it)

 

[GretaGarbo]

The "training" is just to estimate some parameters.

 

[hlsmith]

I will add some more. If you have a sufficient amount of data, the data science protocol is to randomly split your data into three sets to build, say a prediction, model using the training set while tuning the model's hyper-parameters using the validation set. Lastly you apply the built model to third set to tests it based on a chosen error function (eg, missclassification). So you train, validate, and test a model using the sets.