In case you were wondering.
This data is analized through AI, and Big Data. Go look up a tutorial on python for how to do this.
The tutorials will speak about means, correlations, and patterns.
This is a very complex field, so to be short, I will give you its applications instead of how it is done.
Also, to be frank, I don't understand half of it myself.
The best example is weather prediction models, because of how open the whole affair is:
Firstly, data is gathered by hundreds of weather stations, many maintained by hobbyists, and students at universities.
These gather humidity, air pressure, wind direction and speed data, and many others.
This data can be visualized, for wind data in spain: (spain has a lot of wind activity and very interesting as well.)https://www.xcweather.co.uk/ES/observations
Wind data can be correlated with seasons, elevation, and the other factors such as pressure. It could also have to do with how much greenery exists in an area.
Then along comes a Big Data analyst. with a computer, takes all of this data, and spends weeks trying to find correlations.
Such correlations as temperature and presence of trees. Trees keep the area slightly cooler.
So from this correlation we can derive that cities should plant trees on walkways, this is better for people!
This data can also be used to create a model.>At any instance in which "this and that" happens, this other thing will also happen 90% of the time.
When a cold front is observed, certain winds and rains can be predicted in it's wake.
Tornadoes are predicted like this as well.
So how do we jump to predicting the purchase habits of most pregnant women who like fitness?
Enter the silicon revolution.
Your smartphone alone is more powerful than the entirety of computational power of the NASA in the 70s.
Dates back in the 90s were stored as a series of numbers. To save space, the year used 2 numbers alone instead of 4.
Today, I can store dates in their whole string format and nobody gives a shit; such as "The first of October of the year two thousand and thirty-nine".
It would be a stupid idea but I could do it, and if I did go this stupid route, I could parse and use this information with minimal impact to my app.
This is because computers are comically overpowered. Most people subestimate how absurdly powerful computers are, it isn't even funny.
So, couple this capacity to store arbitrary amounts of data, with asburd processing power.
Social media harvests lots of data, even the arbitrary kind, and then people with the job description of Big Data anlists come in and tell the executives, "here, I found a pattern, now gibs me the reward."
This data is analized real time, packaged, and sold to whoever pays more for it under a convenient subscription model.
Imagine as a company that sells shoes, that you know what trends are in vogue, and how many people you can expect to enter your store based on who you appeal to.
Or even better yet, who buys the most shoes!
Apparently, it is this mythical pregnant woman who likes crossfit, for some inane reason.
Nobody knows why, they just know that is how it is. The numbers do not lie.
And AI is a step forwards in this same direction.