Projects
The WhatsApp chat analysis tool serves as an invaluable resource, offering essential statistics, comprehensive timelines, deep content insights, and in-depth sentiment analysis. Its versatility shines through as it adeptly manages both group dynamics and individual conversations. This multifaceted tool empowers users to uncover intricate messaging patterns and delve into the nuanced sentiments woven throughout their conversations, thereby enhancing their understanding and facilitating more informed communication strategies.
The Adventure Works Sales PowerBI Analysis project delves into data from the Adventure Works shop, focusing on Orders, Customers, and Products tables. It encompasses data preprocessing for clean data, implements a Snowflake Schema for efficient connections, and utilizes DAX for measures. The dashboard pages include an Executive Dashboard featuring KPIs and dynamic visuals, a global map for orders, product and customer insights, and profit scenario analysis, providing a comprehensive view of business performance and customer trends.
The Heart Disease Prediction project focuses on using machine learning models and data analysis to predict heart disease presence or absence. It includes data preprocessing, feature engineering, and model fitting with various algorithms like Logistic Regression, SVM, KNN, Decision Tree, Random Forest, and Gradient Boosting. The top-performing model, Random Forest, achieved an 85% accuracy and an F1-score of 0.85. This project provides valuable insights into key predictors of heart disease.
The "AtliQ-Hardware-Sales-Analytics" project involves creating two essential reports: Sales and Finance using Excel. The Sales Report focuses on customer performance and market comparison, aiding businesses in monitoring sales and identifying key performance indicators. The Finance Report creates Profit and Loss reports, supporting financial evaluation, decision-making, and stakeholder communication. Proficiency in ETL, Power Query, DAX, and user-centric report design are essential skills for success in this project.
I embarked on the task of generating real-time raw data using Indian currency notes and leveraged the ImageDataGenerator for image augmentation during the rigorous model training process. The culmination of these efforts yielded remarkable results, with testing accuracy soaring to 90.17% with Resnet-50 and an astounding 95.5% with Yolov8m-cls, across an extensive dataset comprising 900 unseen images. These exceptional findings not only underscore the efficacy of the approach but also lay the groundwork for potential research paper publication, demonstrating the significance and potential impact of this work in the field of computer vision and currency recognition.
