Projects

Replication of "Scale-Free Networks Are Rare"

Replicated the findings of the influential paper "Scale-Free Networks Are Rare" by Anna Deokar and Mark E. J. Newman to critically evaluate the prevalence of scale-free networks in real-world datasets.
Reproduced the statistical analysis conducted in the original paper using a variety of real-world network datasets. Employed rigorous statistical methods to test for the presence of power-law distributions in the degree sequences of networks.
Utilized the powerlaw library to fit power-law distributions and compare them with alternative distributions. Implemented statistical techniques to ensure the robustness of the power-law fitting process, avoiding false positives. Analyzed a diverse set of networks including social, biological, technological, and information networks using the networkx library.
Collected and processed datasets using pandas for data manipulation and cleaning. Utilized numpy for numerical operations and data transformations.
Created comprehensive visualizations using matplotlib to illustrate the degree distributions and the goodness-of-fit for different statistical models. Developed plots to compare the empirical degree distributions with the fitted power-law and alternative models.
Confirmed the findings of the original paper, demonstrating that true scale-free networks are relatively rare in the analyzed datasets. Provided a detailed statistical analysis showing that many networks are better described by distributions other than power-laws.
Tools and Technologies Used:numpy, pandas, matplotlib, networkx, powerlaw

CSU, Chico WREC App Resign

Led the redesign of the university gym (WREC) app to enhance user experience and streamline access to gym facilities.
Managed and directed the redesign project, coordinating with stakeholders and team members to ensure project alignment with user needs.
Implemented one-time login functionality to simplify user access and improve the overall user experience. Enhanced the booking processes for gym facilities, making it more intuitive and efficient for users to reserve equipment and spaces.
Integrated modern design principles and best practices in mobile app development to ensure a seamless and engaging user interface. Utilized Kotlin and Android Studio as the primary development tools to create a more responsive and user-friendly application.
Tools and Technologies Used: Kotlin, Android Studio, Firebase, Figma, XML

Evaluation of ML Classification Methods

To compare the performance of various machine learning classification methods on the MNIST dataset, which consists of 10,000+ handwritten digit samples.
Decision Trees: Basic implementation with depth control and pruning to avoid overfitting.
Random Forests: Utilized an ensemble of decision trees to improve accuracy and robustness, with hyperparameter tuning for the number of trees and depth.
Support Vector Machines (SVM): Evaluated both linear and kernelized (RBF, polynomial) SVMs using Grid Search for hyperparameter optimization (C, gamma, kernel type).
Preprocessed the data with normalization and PCA for dimensionality reduction to improve model performance.
Accuracy, precision, recall, F1-score to provide a comprehensive assessment of each model's performance. Confusion matrix analysis to understand misclassifications. Cross-validation (k-fold) to ensure robustness and generalizability of results.
Employed Grid Search techniques to fine-tune model parameters for optimal performance.
Detailed comparison of model performance, highlighting strengths and weaknesses of each method. Visualizations of training and validation accuracy/loss curves. Analysis of computational efficiency and scalability of each method.
Tools and Libraries: Utilized Python with libraries such as scikit-learn, NumPy, Pandas, and Matplotlib for implementation and visualization.

Analyzing Voting Results using Influence Matrix Technique

Developed a predictive model to forecast election winners based on historical results and current survey data using advanced data mining techniques.
Data Acquisition and Modeling: Collected and processed datasets from publicly available sources, treating them as outcomes of a discrete random process. Utilized these datasets to analyze voting patterns and behaviors over time.
Algorithm Implementation: Designed and planned the implementation of the DeGroot algorithm and maximum-a-posterior (MAP) estimator. Formulated an influence matrix (transition matrix) to capture the dynamics of voter influence and opinion shifts.
Tools and Technologies Used: .NET Framework & C#: Core development environment and programming language for implementing the predictive model. Data Mining Techniques: Applied statistical and data mining methods to analyze and predict election outcomes. Algorithmic Implementation: Developed the influence matrix using DeGroot and MAP estimators to model voter behavior and predict election results.
Published the paper - International Research Journal of Engineering and Technology (IRJET), S.NO:149 Vol 6, Issue 8, Aug 2019

Providing Secure Session through Encryption of Logs for Financial Services

Proposed and developed a framework for a secure and cost-efficient cloud-based log management service tailored for financial transactions.
Framework Design: Conceptualized and designed a robust framework to securely manage log files of bank transactions in a cloud environment.
Encryption Techniques: Utilized a homomorphic encryption algorithm for the secure analysis and storage of log files. Implemented a secret key cryptosystem to ensure the protection of log records during transit and while stored in the database.
Security and Integrity: Addressed and resolved security and integrity issues in log maintenance by implementing an encrypted log management system. Ensured data confidentiality and integrity through rigorous encryption protocols and secure logging mechanisms.
Prototype Development: Developed a proof-of-concept prototype to demonstrate the feasibility and effectiveness of the proposed secure logging approach. Maintained and managed encrypted log records, showcasing the practical application of the system
Published the paper - International Research Journal of Engineering and Technology (IRJET), S.NO:53 Vol 6, Issue 1, Jan 2019
Tools used: .NET framework, C# , Front End Asp.Net