Student Depression Prediction
An academic machine-learning classification project.
- Logistic Regression reached 83.66% accuracy
Overview
An academic machine-learning project framed as a binary classification task. I prepared the dataset, trained three classic models and compared them to find the best balance of precision and recall.
This was an academic classification exercise. It is not a medical or diagnostic tool, and its results make no clinical claims.
Dataset
- 27,901
- 18
- 80 / 20
- Binary classification
Implementation
From raw data to a compared set of models.
Preprocess
Cleaned the data and handled missing values.
Encode
Applied label encoding and one-hot encoding.
Split
80/20 train–test split.
Train and compare
Logistic Regression, Decision Tree and K-Nearest Neighbors.
Result
- Logistic Regression
- 83.66%
Logistic Regression was identified as the best-performing model, offering the most balanced precision and recall of the three.
Future improvements
- Try Random Forest
- Try Gradient Boosting
- Further feature engineering
