TL;DR, predicting poker hands.
Given the sequence of 5 'community' cards, drawn from a standard deck of cards, what hand is most likely present with at least one of the players in the game.
Diligence is the mother of good luck.
- Benjamin Franklin
Texas Hold Em is played by dealing each player 2 cards (face down), called the hole cards, and dealing 5 community cards (face up), on the table.
The player makes a poker hand using any combination of the 3 cards dealt to them, and the 5 cards on the table.
The player with the strongest hands wins. Most commonly accepted ranking of hands, strongest to weekest :
| Rank | Hand | Description |
|---|---|---|
| 0 | Royal Flush | A K Q J 10 all of the same suit |
| 1 | Straight Flush | Any 5 cards of the same suit, in sequence |
| 2 | Four of a Kind | 4 cards of the same rank, like, 4 4 4 4 |
| 3 | Full House | A 3 of a kind, and a pair, of different ranks |
| 4 | Flush | Any 5 cards of the same suit |
| 5 | Straight | Any 5 cards in sequence |
| 6 | Three of a Kind | Any 3 cards of the same rank |
| 7 | Two Pair | Any 2 pairs of cards |
| 8 | One Pair | Any 2 cards of the same rank |
| 9 | High Card | Highest Ranked card in hand |
A K Q J 10 9 8 7 6 5 4 3 2 1
This data was acquired from UCI's Machine Learning repository. The data comes already split into training and testing data.
Find it here.
Each bar represents the frequency of the corresponding hand as specified in the dataset description.
NOTE: The hands' class labels are in the reverse order of their strength, i.e, 0 is the weakest hand.
Dua, D. and Karra Taniskidou, E. (2017). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science.
| Model | Accuracy | Plot |
|---|---|---|
| Linear Regression | 42% |  |
| SVM | 58% |  |
| Adaboost | 49% |  |
| Output Code Classifier | 61% |  |
| Random Forest | 56% |  |
| Artificial Neural Network | 45% |  |
| Deep Neural Network | 87% |  |
| Multi Layer Perceptron | 97% 🤯 |  |
The Multi-layer Perceptron is clearly the best model for the dataset in hand. Here's the confusion matrix to provide us with some more insight into how accurate this model really was with its predictions.
 |
Aditi Srinivas |
 |
Avinash Shenoy |