# BlackJack Smart Effect of Removal ML — Data

来源：https://www.kaggle.com/competitions/black-jack-smart-effect-of-removal-ml/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

# Dataset Description

## Overview
The dataset represents Blackjack deck composition scenarios with their corresponding Expected Values (EV). Each row shows how many cards of each value (1-10) were removed from a 6-deck shoe and, in the training set, the resulting EV of the remaining deck.

## Files
* **train.csv** - 20,500 scenarios with known EV values
* **test.csv** - 8,655 scenarios where EV needs to be predicted
* **sample_submission.csv** - example submission format

## Columns
* `id` - Unique identifier for each scenario
  * Training set: 0 to 20,499
  * Test set: 20,500 to 29,154
* `1` through `10` - Number of cards removed for each card value 
  * Range: 0 to max cards in 6 decks for that value
  * Example: Column `1` represents Aces removed (0-24 possible)
  * Example: Column `10` represents 10s/face cards removed (0-96 possible)
* `ev` (training only) - Expected Value of the remaining deck
  * Range: Typically between -0.05 and +0.05
  * Negative values indicate house advantage
  * Positive values indicate player advantage

## Common Terms
* **EV** - Expected Value: the average amount a player can expect to win or lose per hand
* **EOR** - Effect of Removal: how removing specific cards affects the deck's EV
* **CDCA** - Casino Domain Combinatorial Analysis: tool used to generate the data
* **Shoe** - Container holding multiple decks (6 in this case)

## Submission Format
Predictions should be made for each scenario in test.csv, containing:
* `id` - Row identifier (20,500 to 29,154)
* `ev` - Predicted Expected Value

Example submission: