Modeling typing performance: Insights from a novel dataset and minimal attributes
Abhinav Chaitanya, Aman Gupta, Abhilasha Sharma · 2025
Typing is a vital skill in the present age that has a major impact on various activities of daily life. Hence, it is quite important to analyze the various factors that impact typing performance. A lot of research has been accomplished in areas of user authentication, behavioral patterns, keyboard ergonomics, and the influence of cognitive factors like memory, fatigue, and attention. Much of this research has been around keystroke dynamics. This paper aims to identify the various patterns in typing performance particularly for people taking part in online typing tests by providing a new dataset of 15,003 typing tests given by 22 users on the Monkeytype platform. The dataset provides a unique opportunity to analyze typing performance patterns. The research investigates the feature importance of the key features that influence typing speed and examines their predictive power using different machine learning algorithms such as Random Forest, ExtraTrees, XGBoost, and Catboost as well as neural network models. Via a comparative evaluation of these models, we aim to identify the best approaches for WPM prediction, followed by statistical testing for hypotheses indicating performance variations in typing sessions. Further, the correlation and tradeoff between accuracy and consistency are explored.