Machine Learning and statistics

justnum

justnum

Billiards Improvement Research Projects Associate
Silver Member
There is a lot of tournament data and a lot of statistics people here.

Has anyone tried to apply machine learning to train datasets?

Since Fargorate is taking the world by storm, I am thinking of training a network to make pool predictions and ultimately calculate the odds for the outcomes of different players matching up.

Fargorate is for ranking. I am interested in regression and estimation.
I believe jumps in skill level are discrete.
 
I haven't done this for pool but think it's a good idea. I suggest python and tensorflow. Google will give you a free colab notebook to train your network and share your data. Check out the Kaggle tutorial on predicting deaths on the titanic, the same analyis could be used to try to predict pool outcomes.
 
Black-Balled said:
I thought you were taking a break from Azb for a couple months?
Click to expand...
programming, and data science doesnt count, I will just post results of who should've beaten who and when. After my project is finished.
 
Black-Balled said:
Before or after that 50 greatest safes video?
Click to expand...
I plan to let the computer sort through the vids for the images I want.
there is no way I am manually editing those videos.

Two computer projects maybe I will take on a third, got any ideas.
I used to complete these things overnight. I've been busy with my paid job
 
1200px-Plankton_and_Karen.svg.png
 
I.think it would be easy (but time consuming) to build a computer vision model that predicts run out in 9-ball after the break. It would require spending time to capture1000 images of spreads that lead to runouts and 1000 images of spreads that did not. You could load the model in a mobile app to predict run out based on the camera input.

There is not any utility to this except I would feel better when I screw up and don't run out (that the model would confirm my view that a runout was difficult).
 
The hardest part here will be finding the data. Which data set(s) do you want to use and how do you plan on getting them?
 
skip100 said:
The hardest part here will be finding the data. Which data set(s) do you want to use and how do you plan on getting them?
Click to expand...
A response to your question would require thought.

Numz just posts whatever musings flicker through his thoughts, with no follow-through or further actions.
 
skip100 said:
The hardest part here will be finding the data. Which data set(s) do you want to use and how do you plan on getting them?
Click to expand...
I've been watching the AZ player pool, and consider players from the same country and track from title wins of country.

Filipino dominance can be proven statistically.
 
Where is the pool data? Has anyone undertaken a study to determine the degree young people are being attracted to pool? Until we base opinions on objective data, we are simply bumping our gums.
 
justnum said:
I've been watching the AZ player pool, and consider players from the same country and track from title wins of country.

Filipino dominance can be proven statistically.
Click to expand...
A machine learning model will require a clean, labeled data set with match results for hundreds or thousands of players. This kind of database is the actual secret sauce for FargoRate, not the specifics of its statistical model.

Good luck in your efforts.
 
skip100 said:
A machine learning model will require a clean, labeled data set with match results for hundreds or thousands of players. This kind of database is the actual secret sauce for FargoRate, not the specifics of its statistical model.

Good luck in your efforts.
Click to expand...
I know there are gamblers here who play the stocks.
Your clean data may be fudged because players are known to "fake it"

Whats the famous line? A pro can make more money in the practice room that winning the tournament.

The reaction during a match when it happens is how you know.

Filipinos dont need to win prizes, taxes will want their cut. And the conversion from dollar to peso is through the roof.
assuming equal desire to place high breaks your data easily
 
justnum said:
I know there are gamblers here who play the stocks.
Your clean data may be fudged because players are known to "fake it"

Whats the famous line? A pro can make more money in the practice room that winning the tournament.

The reaction during a match when it happens is how you know.

Filipinos dont need to win prizes, taxes will want their cut. And the conversion from dollar to peso is through the roof.
assuming equal desire to place high breaks your data easily
Click to expand...
How will you overcome 'fake it' and have a result that is true?

You can't tell Jack shit from interpretation of participants' post match reactions.

The europeans have surpassed the pinoys in pool, as have the Chinese, sorry to say.
 
Black-Balled said:
How will you overcome 'fake it' and have a result that is true?

You can't tell Jack shit from interpretation of participants' post match reactions.

The europeans have surpassed the pinoys in pool, as have the Chinese, sorry to say.
Click to expand...
Thats classified information.
 
You must log in or register to reply here.
Back
Top