Overview

A deep learning model analyzes user sentiment from DANA app reviews, addressing challenges like imbalanced data, overfitting, and computational limitations. Using SMOTE oversampling, the dataset was balanced, while Dropout and Batch Normalization improved model generalization.

Cloud based GPU acceleration optimized training time, and extensive hyperparameter tuning refined model performance. These techniques ensure an accurate and scalable sentiment analysis system, providing valuable insights into user feedback.

Found out more on my Github.