Chess Position Evaluations

The dataset includes 513 million chess position evaluations performed by the Stockfish engine for analysis on the Lichess platform.

Source: Kaggle - Chess Position Evaluations

Size: 59.66 GB

Loading Example

1. Download the evals.csv file from Kaggle.

2. Create a table in YDB using one of the following methods:

   Embedded UI

   YDB CLI

   For more information on Embedded UI.

   CREATE TABLE `evals` (
       `fen` Text NOT NULL,
       `line` Text NOT NULL,
       `depth` Uint64,
       `knodes` Uint64,
       `cp` Double,
       `mate` Double,
       PRIMARY KEY (`fen`, `line`)
   )
   WITH (
       STORE = COLUMN,
       UNIFORM_PARTITIONS = 50
   );
   

   ydb sql -s \
   'CREATE TABLE `evals` (
       `fen` Text NOT NULL,
       `line` Text NOT NULL,
       `depth` Uint64,
       `knodes` Uint64,
       `cp` Double,
       `mate` Double,
       PRIMARY KEY (`fen`, `line`)
   )
   WITH (
       STORE = COLUMN,
       UNIFORM_PARTITIONS = 50
   );'
   

3. Execute the import command:

   ydb import file csv --header --null-value "" --path evals evals.csv
   

Analytical Query Example

Identify positions with the highest number of moves analyzed by the Stockfish engine:

SELECT
    fen,
    MAX(depth) AS max_depth,
    SUM(knodes) AS total_knodes
FROM evals
GROUP BY fen
ORDER BY max_depth DESC
LIMIT 10;

ydb sql -s \
'SELECT
    fen,
    MAX(depth) AS max_depth,
    SUM(knodes) AS total_knodes
FROM evals
GROUP BY fen
ORDER BY max_depth DESC
LIMIT 10;'

This query performs the following actions:

• Finds positions (represented in FEN format) with the maximum analysis depth.
• Sums the number of analyzed nodes (knodes) for each position.
• Sorts results by maximum analysis depth in descending order.
• Outputs the top 10 positions with the highest analysis depth.