GitHub Malicious Accounts Detection
Fall 2023
Objectives:
Distinguish malicious accounts by studying the patterns in accounts creation on Githhub using R or Python
Actions, Results & Insights:
- Performed Exploratory Data Analysis to analyze and visualize over 3 thousand labeled data to extract
insights from it
- Developed binary classifier using the Random Forest algorithm to detect fraudulent
accounts creation and achieved an accuracy of more than 97%.
- Statical analysis revealed 445 fraudulent accounts and 3399 non-malicious
Tools & Library used: Python, R, Excel, Matplotlib, Pandas, Numpy, Collections, Itertools
Modeling Credit Risk Project at Equifax
Spring 2023
- Developed an 89% accurate binary logistical model able to identify consumers' likelihood of default
- Delivered results in scores to consumers through an interactive web application
Automating the Categorization of Issue Tickets at Southern Company
Fall 2022
- Used Natural Language Processing to categorize over 10,000 service tickets
- Created a model with 80% accuracy combining principal component analysis and support vector machine
- Reduced labor by 50%
Lean Process Improvement for XPO Logistics Distribution Center
Spring 2020
Objectives:
Implement a process to transition from a floor loading to a palletized system. This new process would reduce safety risks and minimize waste in the form of non-value-added material handling. It should also result in a reduction in wasted space and damages, improve item throughput efficiency, and reduce labor hours caused by the current floor loaded shipping system.
Actions:
- Performed alternative analysis for proposed pick row redesigns.
- Performed a relationship analysis to determine the impact of the proposed pick row designs with the project requirements.
- Accessed the warehouse’s Kronos timekeeping system to establish a baseline for average cases moved per week, labor hours, productivity levels, and cost per case.
- Requested the creation of a case strength index from packaging/material engineers.
- Mocked the new pick row design and implemented the new design layout
- Performed measurements and comparative analysis with current and past Kronos data.
Results and Insights
- Significant reduction in the cost per case from $0.307 in quarter one to $0.272 in quarter three.
- Company will see an expected savings of $274,074.30 for the first year and an additional $82,589.67 in the following years from a reduction in labor cost.
- KPI’s for productivity resulted in an increase in cases per row of 9.71 compared to the mean of 9.54, a decrease in daily case per hour of 302.35 compared to the mean of 305.70, and a decrease in rows visited per hour of 31.14 compared to the mean of 32.05
Tools used: MS Word, Excel, PowerPoint, Filmora9, Google Applications, Red Prairie WMS, Kronos Timekeeping System