Graduation Year

2024

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Mathematics and Statistics

Major Professor

Chris P. Tsokos, Ph.D.

Committee Member

Kandethody Ramachandran, Ph.D.

Committee Member

Lu Lu, Ph.D.

Committee Member

Sajeev Varki, Ph.D.

Keywords

Finance, Engineering, Optimization, Analytical

Abstract

In recent years, social media has emerged as a highly appealing investment sector due to its substantial returns. Our research focuses on creating a predictive model for the financial performance of social media companies. We gathered data from five companies in the S&P 500: Meta, Etsy, Match, Pinterest, and Alphabet. We developed a non-linear, data-driven stochastic analytical model that employs real data, incorporating six financial indicators and four economic indicators, to forecast the weekly closing prices of these social media companies with an accuracy of 94.7%. Additionally, we ranked the statistically significant indicators and their interactions based on their percentage contributions to the returns of social media stocks. Next, we utilized Response Surface Methodology to optimize the weekly closing prices of the five social media stocks from the S&P 500: Meta, Etsy, Match, Pinterest, and Alphabet. We calculated both the 95% Confidence Interval and the 95% Prediction Interval for the projected weekly closing prices of these stocks. Furthermore, we identified the optimal levels of risk factors that maximized returns with a high degree of accuracy. Another area of our research is the reliability of Hard Disk Drives (HDD). HDD failures can have significant consequences for both individuals and organizations. The drawbacks of HDD failures highlight the importance of data reliability and the need for effective backup and recovery strategies. We selected 5 models of the Seagate company, comprising a total of about 2000 HDD. Our first task was to perform parametric analysis and obtain the respective reliability function of the 5 models. We were able to perform parametric analysis on 4 of the models, and determined their reliability function. The 4 models were ranked based on their reliability estimates. ST4000DM000 was the model with the highest reliability estimate. There was one model, ST4000DX000 for which we couldn’t conduct parametric analysis. In the absence of parametric analysis, we performed non-parametric analysis of HDD failure times using Kaplan-Meier and Kernel Density Estimation, which are two widely utilized methods. In Kernel Density Estimation, the choice of kernel and bandwidth is crucial for the analysis. Our study found that Kernel Density Estimation is more powerful, robust, and effective compared to Kaplan-Meier. Last, we presented a modern analytical approach that utilizes the Reliability Index (RI) to monitor and evaluate the performance of our best Seagate Model in terms of reliability, ST4000DM000. We analyzed the failure times of approximately 2,000 HDDs of this model, which is widely regarded as one of the most reliable. We primarily introduced two methods: the Reliability Indicator (RI) and the Stochastic Intensity Function (SIF). Our analytical approach, incorporating both the SIF and RI, serves as a contemporary method for tracking and assessing reliability over time and is applicable to any disk drive.

Download

Share

COinS