Player activity data: an individual planning report

My individual plan for a group project on player data from a research game server. I joined two tables, explored the data in three charts and proposed a model.

Role
Author of the individual report (coursework)
Period
Fall 2024
Stack
Python, pandas, Altair, scikit-learn
Evidence
Published
Question
Which kinds of players contribute the most play time to the research server?
Data
Course data from a research game server: 196 players and 1,535 play sessions.
Method
I joined the sessions to the players and calculated the total session minutes for each player as the response. I made three exploratory charts.
ResultPublished
A proposal for a k-nearest-neighbours regression with scaled inputs, a 75/25 split and five-fold cross-validation.

A plan, not a fitted model. This page gives no model result. The final report of the group took a different approach and is not shown here.

196Players
1,535Play sessions
3Exploratory charts
5Cross-validation folds (proposed)

What the report contains

  1. A description of the two tables and their variables.
  2. The join of sessions to players, and the total session minutes for each player.
  3. Three exploratory charts of play time.
  4. The proposed method: the split, the scaling, the model and the cross-validation.

Why this method

The response is a number, so the task is a regression. A k-nearest-neighbours model makes no assumption about the shape of the relation. It needs scaled inputs, because it uses distances.

Cross-validation on the training part selects the number of neighbours. The test part stays unused until the end.