eli5: What’s the difference between Data Mining and Machine Learning?

173 views

[ad_1]

To me, both look the same. Both are used to find patterns in data or to learn something from it so where’s the difference?

In: Technology
[ad_2]

Data mining: A human looking for something in a large dataset

Machine learning: Computer programs (AIs) that learn from a large dataset to produce similar, original results.

Data mining is trying to understand how human work to predict their action. It’s using information gathered about people to try and understand the pattern so you can make advertisment that will work better. Example: By looking at people’s browser history, they realize that people that buy a lot of beer on a day before a big football match also usually buy charcoal and meat. So they decide to make a “barbecue” package when a big football match is coming.

Machine learning is a program learning from its mistake. The program tries to do something, fail then try to change its approach. try again, fail again, change approach. Try again, fail again, change approach. Try again, succeed for once, expand on this method. Try again using previous method but slightly tweaked, no longer work, tweak differently. Repeat, repeat, repeat. It’s doing the (almost) same thing over and over and over and over again observing things changing. Pick what works best and try to see if that idea can be tweaked to do better. Rinse and repeat.

Data mining is searching for similarities to somewhat predict the future (mostly used to predict people buying pattern)

Machine learning is letting a program learn for itself how best do a job.

They are related, but not all data mining is ML and not all ML is data mining.

Data Mining is a wide field that involves finding, analyzing and correlating data. You can do this using ML, but you could also do this by hand, using statistical methods, or non-ML methods. All you’re doing is looking at data and finding as many useful patterns as you need.

Machine learning is also quite a wide field really that loosely is defined as “Computers teach themselves to do stuff”. Among the ‘stuff’ is pattern extraction and recognition but it can also be “learn how to most effectively move from A to B when there are holes in the floor, the ability to create original pieces of art, or get really good at playing games.