I created an ASL Translator using CNN for image classification and real-time hand detection with OpenCV and MediaPipe. Used a dataset of 174,474 ASL alphabet images, achieving 99.19% Test Accuracy and 99.96% Train Accuracy. Built user-friendly web apps with Streamlit and Flask, enabling image uploads and real-time webcam translations to enhance communication between deaf and hearing individuals.
This project's goal is to utilize an API to scrape subreddits and build a natural language processing model for predicting post origins. I implemented a pipeline with TF-IDF Vectorizer and Logistic Regression, optimized with Grid Search, achieving over 90% prediction accuracy.
In a six-hour hackathon, I developed an image classification model for hotdog detection. Additionally, I crafted a user-friendly Streamlit app enabling picture uploads for real-time hotdog predictions.
My team and I embarked on a project where we designed and implemented a sophisticated Random Forest regression model aimed at predicting flight ticket prices. Leveraging advanced techniques, such as feature importance analysis, we successfully pinpointed crucial predictors within the dataset. Impressively, our model demonstrated exceptional performance, achieving an accuracy rate surpassing 97% with a low Root Mean Square Error (RMSE) of 42.75.
In this project I analyzed housing data in Ames, Iowa and created a Lasso Regression model that achieved 88% precision in its predictions, predicting sale price within an approximate margin of $21,000.