Data science and supervised learning

Supervised learning explanation

Supervised learning is a technique. You can say that based on past data, there are many attributes related to the data set. You also have something called a tag. Supervised learning creates the perception of objects, and by tagging objects helps identify not only the object but also its future variability.

Learning as a child

So, for you, it seems that you are a child, learning to recognize different kinds of fruits, for example. You look at the fruit intuitively, you know what Apple is like. You form a psychological perception around it. Someone tells you that anything that looks like this shape is an apple. Other fruits are similar, such as bananas, oranges, etc. Therefore, you have learned this visual perception since childhood, and the help that another person gets from others tells you that your visual perception is an apple. This is called supervised learning.

Supervised learning input

Your perception has an input function, more around the color, shape and structure of the fruit, and someone tells you that this kind of thing is called apple. Therefore, the combination of the two, the machine learning model trains itself. For a while, regardless of the shape, color and texture of the different types of apples, you can be sure that this is an apple. So no matter how different the tricks you have made, no matter how the future of nature and the emergence of new varieties of apples, your views on identifying apples are very strong, because someone has trained you. This is usually what happens in the machine learning model.

Need training and accuracy

You train yourself with a lot of input data about any given object, and based on which you have a tag, this tag tells you that this is an apple. Remember, because we are training someone about the object, your data should be 100% correct whenever you plan a dataset for supervising machine learning algorithms. Even if you missed a 10% data set and you think the label is wrong, you will have 10% as an output error. Simply put, your model is as good as your data.

In summary

There are many algorithms that can build supervised learning. Make sure you understand them during your data science training. For example, if you build a classifier for a fruit, then the label that will be applied - this is a banana, which is an apple, which is orange, based on an example of a classifier that displays bananas, apples, and oranges, respectively.

---

Orignal From: Data science and supervised learning