This project revolves around tackling the complex mystery presented by the VAST Challenge in January 2014. The enigmatic disappearance of GAStech employees during their company's IPO celebration, with suspicions pointing towards the Protectors of Kronos (POK), creates a compelling backdrop. GAStech's secretive geospatial tracking software, concealed within company vehicles, serves as a crucial data source, despite missing data for the critical day of the disappearance. Law enforcement has also provided access to two weeks of credit and debit card transactions and loyalty card data for local GAStech employees.
As a data science expert aiding the investigation, our mission is to harness the power of data visualization and analytics to decipher this information. We must unravel the complexities arising from incomplete and inconsistent data to identify suspicious behavioral patterns and prioritize leads related to the missing personnel. This project embarks on a journey through data exploration and visualization to shed light on the enigma of the missing GAStech employees and assist law enforcement in their quest for answers.
The primary dataset used in the analysis is the GPS data, which was combined with the car assignment data to create a comprehensive data frame. By merging the "gps.csv" file with the "car_assignment.csv" file, a cohesive dataset was created. This integrated data frame provides essential information such as timestamps, longitude, latitude, driver names, car IDs, and employment details. This dataset serves as the foundation for mapping employee locations at specific points in time, enabling the visualization of their movements two weeks prior to the incident.
Given that the GPS data lacks explicit stop information, the "stop data" fills this gap. A specialized function was developed to iterate through the main data frame, calculating time intervals between rows. When the time interval exceeded five minutes, a stop flag was incorporated into the dataset, enabling a comprehensive understanding of where employees visited during their days.
Card transaction data provide insights into employees' financial activities. The integration of credit card and loyalty transaction data into a unified data frame facilitated a comprehensive analysis of financial behavior. Transaction-type flag was added to distinguish between credit card and loyalty card transactions, enabling a nuanced exploration of spending patterns. This combined dataset is created to unravel potential correlations between employee behavior and financial activities, further strengthening the analytical framework.
These Python files collectively laid the base groundwork for the dashboard development using Plotly Dash. The processed merging, cleaning, and categoriztion of diverse data sources created in a unified dataset ready for analysis. This integrated dataset helps this project to examine and bring together employee geospatial-temporal patterns with credit card transaction records. The dashboard, a visual interface powered by the prepared data, is used as an important tool to analyze employee movements, investigate anomalies, and establish accuracy in credit card transactions.
The visualization phase of the project used Plotly Dash to provide an interactive and comprehensive representation of the cleaned and merged data. The integration of Plotly Dash component tabs facilitated a structured and user-friendly interface for exploring the diverse facets of employee movement and behavior.
The first tab, dedicated to "Truck Driver Movement," addresses the unique challenge of identifying truck drivers for various routes and dates. The dynamic graph displayed on this tab enables users to select specific routes and dates. By correlating the GPS data with the truck routes, this visualization aids in pinpoinƟng the responsible truck driver for each route on distinct occasions. This insight provides clarity into truck driver assignments and route allocation, contributing to a comprehensive understanding of the geospatial-temporal patterns associated with truck drivers.
The second tab introduced an intuitive dropdown menu enabling the selection of distinct employee categories. An additional time dropdown facilitated the isolation of movements during specific periods of the day. The chosen parameters dynamically updated the graph, allowing users to visualize employee movement patterns based on their category and chosen time. Complementing the graph, a table provided details, enhancing the understanding of employee trajectories. A weekend radio button was added, to switch between weekday and weekend views. This option aimed to unveil disparities in general employee movement behaviors between these timeframes.
In the "Individual Employee Movement" tab, users can select an employee's name, along with a date and time range. The resulting graph provides a detailed trajectory of the selected employee's movement during the specified period. A supplementary table showing card transaction data is positioned under the graph. This integration facilitates a comprehensive analysis, enabling users to discern potential anomalies or irregularities in an individual employee's movement patterns and associated financial activities. By comparing movement data with card transactions, this tab supports the identification of suspicious behaviors and aids in confirming the legitimacy of transactions
Anomalies from the graphed general movement of the employee have shown a group of individuals who needs further investigation: Loreto Bodrogi, Isia Vann, Minke Mies, and Hennie Osvaldo – all of whom hold positions within GAStech's security division. Intriguingly, their activities have been notably conspicuous during the midnight to early morning hours on specific dates: January 7th, 9th, 11th, and 14th. What sets these instances apart is the fact that their GPS locations seem to converge at the residences of prominent executives, namely Ada Campo Corrente, Orhan Strum, Willem Vasco Pais, and Ingrid Barranco. Such patterns raise suspicions, suggesting potential ulterior motives. Given the context of a kidnapping investigation in this project, it is possible that these security personnel could be linked to the abduction. Their seemingly suspicious actions and location near executive houses could indicate a deeper involvement, possibly implicaƟng them as suspects or, at the very least, individuals of interest in a kidnapping scenario.
The name mentioned above often visits these 5 mysterious places where there is no detail of what location it is. They generally visit these places during the lunch break and there are no transactions on their card being charged during their visit to these places. Again the names are similar to the security member that spies on the executive houses but Isia Vann is not included here, instead, there’s Inga Ferro who is also a member of the security team. The table below describes their visiting dates and area
On January 13th night Lucas’s credit card was charged $1000 at 19:20, although his GPS location showed that he was not present at the auto supply establishment during that time; his whereabouts were instead traced to Ouzeri Elian. Several potential explanations surface, each with its own complexities. Furthermore, after the suspicious activities done at night by the security team, further investigation focused on those security members. One theory points to potential credit card theft by Loreto Bodrogi, given his proximity to Ouzeri Elian that night and move to near his home which is coincidentally near from Frydos Autosupply n' More.
An alternative hypothesis implicates Minke Mines, whose GPS location more accurately aligns with Frydos Autosupply n' More. Yet, no corresponding transaction from Minke Mies corroborates this theory. This hypothesis also creates suspicion towards Henk Mies the truck driver, as they shared family name and with a big purchase of $1000, it might be items that need to be transported by truck. However further investigation showed that Henk never drives past/near the shops, thus he is out of suspicion.
Possible credit card theft happened supported with the fact that Lucas's credit card sees no further usage following this event, limited only to his loyalty point utilization up to the 16th Jan.
