Implemented 2 different similarity search methods Sentence2Vec (Sentence2Vec) and Bidirectional Encoder Representations from Transformers (BERT) using two datasets taken from Kaggle. The models were trained on the training data, fine-tuned using validation data, and assessed on the test data to compare the performance between the two methods. Our findings highlight the strengths and limitations of both models, providing insights into their effectiveness in similarity search tasks. The results demonstrate that while BERT's contextual embeddings generally lead to better performance for complex datasets, Sent2Vec provides a computationally efficient alternative with comparable performance on simpler datasets.
Problem Statement: Predicting the user’s next location
Dataset: Trip Legs generated in task 2
Solution: Use the Triplegs generated in task 2 as input into an LSTM model to predict a user's next location.