Machine Learning Prediction Model

1Project overview

An end-to-end supervised machine-learning workflow in R, from raw multimodal data to tested predictions.

2Objective

Train supervised machine-learning models that use workload, recovery, and cognitive data to predict game readiness and injury risk.

3My role

• Cleaned and prepared a 15,420-row multimodal dataset in R.
• Engineered workload, recovery, and cognitive features as model inputs.
• Split the data into training and test sets.
• Evaluating predictions to compare models.
• Trained separate models by sex to address potential prediction bias.

4Engineering approach

1. Data preparation

   Cleaned and structured a 15,420-row multimodal dataset.

2. Feature engineering

   Built workload, recovery, and cognitive features as model inputs.

3. Training and testing

   Held out a test set so model performance is judged on data the model has not seen.

4. Bias control

   Trained separate models by sex rather than one pooled model.

5Analysis and visuals

1. 15,420 rowsmultimodal dataset
2. Clean & engineerworkload, recovery, cognitive features
3. Splittraining and test sets
4. Trainsupervised models in R, one per sex
5. Evaluatecompare predictions on test data

6Results and impact

• 15,420rows prepared
• 2sex-specific models

An end-to-end machine-learning workflow, from raw data to tested predictions, with methods that transfer to manufacturing uses such as quality and downtime prediction.

7Tools and technologies

• R
• Supervised learning
• Feature engineering
• Train/test split
• Exploratory data analysis