Justin Saddlemyer/

Workspace Hackathon - Predict the length of chess matches (DO NOT DELETE)


Workspace Hackathon - Predict the length of chess matches


You are working for an online chess service and the marketing team has asked you to create a dynamic system to determine the number of advertisements to show players based on the expected length of a game. If a game is predicted to be short, they want to limit the number of advertisements they show. However, if the game is predicted to be long, they can confidently insert more advertisements.


To help you, the marketing team has provided you with a dataset of 20,000 chess matches to train your model (available as chess.csv). You are allowed to use the opening_length, but cannot use more than the first 10 moves of the game from the moves column. Your target variable is the turns column.

Submission requirements

Your goal is to create a report summarizing your results for the marketing team.

  • You will need to explain how well you can predict the length of matches in terms that they can understand.
  • You should outline your plans for improving the model and implementing it in production.
  • Finally, you should provide suggestions for how the team can use your model to serve advertisements to players.
# Import the tidyverse

# Load the csv as a data frame and preview it
matches <- read.csv("chess.csv")
  • AI Chat
  • Code