# What can one tell a school kid about statistics and machine learning?

Next week we have an intern from a local school in the house. The concept behind his short internship is to get an idea how the real world works and what certain jobs deal with, how the daily work looks like etc.

Now I wondered, what one can tell/show/demonstrate such a young kid about Statistics and Machine Learning so that he/she

• gets the basic idea of this areas
• is getting enthusiastic (assuming that the kid's prior is not too heavily weighted in favor of other interests)
• wont forget about it the next day

I am primarily looking for sticking images, examples for demonstration etc..

The kid's background:

• 15-16 years old
• basic math concepts are known (what is a graph, rule of three, what is a variable (mathematically, not statistically speaking))

Since the hard part of this question is to explain your area to one without any background knowledge, this question might as well serve as reference for chats with relatives and friends.

Sidenote: I skipped the description of my job intentionally, so that this question is not too specific, this question is about the topic of this site in general.

I do research in computer vision and machine learning, when people ask me about machine learning, I like to mention how the 1 million Netflix challenge/prize is a great example of machine learning (stating the problem and input output), and the 3 million ongoing health prize challenge.

The Kinect uses random forests (very popular machine learning approach) and it's really simple to explain.

The website Kaggle.com hosts machine learning competitions, I WISH there was such website when I was a kid. Maybe you want to show it to him.

I would use some drawings to introduce concepts of classification and over fitting.

• the kinect example is great ... can create a link to the youngster ;). Can you recommend some documents explaining the application of RF in Kinect ? – steffen Feb 9 '12 at 20:27
• @steffen: the RF algorithm used in the Kinect is describe in details in this paper (received best paper award at CVPR 2011) - I couldn't find any simplified explanation of it online, but I'm willing to write one tonight and post it somewhere here on Stack Exchange. If you think it's a good idea, let me know an appropriate place in SE where I can post an illustration of the algorithm. – Roronoa Zoro Feb 9 '12 at 20:53
• thank you. I hope/think I can handle the paper, but thank you very much for the offer ! – steffen Feb 9 '12 at 21:03

Explain PageRank. It's got a nice intuitive explanation via a random surfer, and its effects in the real world are immediately obvious.

• +1: i do the same. But then you have to spoil it saying that in the real world it was only used at the very beginning and google's success was more about its ability to scale (i.e. engineering) than math ideas. – user603 Feb 9 '12 at 10:23
• I like the Pagerank example, but to make the pitch more engaging i would rather point out to future challenges than past achievements. To stay on around google, i would mention driverless cars. – user603 Feb 9 '12 at 10:30

If you want to make it somewhat more fun for kids to think about, here's a pretty simple example: http://www.extremetech.com/extreme/88610-rybka-the-worlds-best-chess-engine-outlawed-and-disqualified

When I was first learning about machine learning, it was those kinds of examples that worked the best for me. I think I've heard of people applying the same concepts to some of the RTS games out there now, like Starcraft and Warcraft.