ScholarWorks (California State University)
Not a member yet
87523 research outputs found
Sort by
California affordable housing initiatives and its impact on middle-income homeownership
No full textThis systematic literature review examines California's pro-housing reforms, including Senate Bills 8 and 9, inclusionary zoning, and Accessory Dwelling Unit (ADU) reforms, and their effectiveness in helping middle- and low-income households transition from renting to owning. An analysis of 26 peer-reviewed articles evaluates policy impacts on affordability and ownership. Findings show that the effects on affordability and ownership are limited in relation to increased housing development in some areas of the state. ADUs tend to benefit wealthier homeowners, inclusionary zoning cannot meet demand, and short-term rentals (STRs) reduce long-term rental availability in high-demand neighborhoods. Future research should evaluate the long-term policy impacts across various groups, including those in urban, suburban, and rural contexts
Aligning Strategy with Structure: A Proposal for Organizational Effectiveness in the Department of Public Social Services
No full textPublic sector organizations, such as the Department of Public Social Services (DPSS), face increasing pressure to enhance service delivery and improve organizational efficiency amid diverse and evolving community needs. Structural misalignment with organizational strategy has led to inefficiencies and decreased responsiveness in large agencies, pinpointing the importance of exploring effective alignment approaches. This study utilizes systematic literature review grounded in contingency theory and strategic frameworks to examine how strategic alignment between structure and organizational strategy can improve performance. It also addresses how structural configurations of centralization and decentralization correspond with strategic alignment like defender and prospector within DPSS. The findings reveal that tailored alignment between structure and strategy improves performance by enhancing efficiency, innovation, and responsiveness across divisions with differing strategic requirements. These insights offer valuable implications for policy and practice, advocating for differentiated organizational designs in public agencies and emphasizing the necessity of strategic flexibility in dynamic service environments. The study also identifies avenues for future research on structural reform and performance measurement
Diverse mechanisms of enzymatic browning: a tri-species comparative study of Malus domestica, Olea europaea L., and Solanum tuberosum L.
No full textThe purpose of this review is to synthesize research findings from three published studies that investigated enzymatic browning. The first centers on the characterization of PPOs from Golden Delicious apple, where N-terminal microsequencing identifies two distinct isoforms specific to this cultivar. The second study presents a detailed protocol for extracting PPO from olive that successfully removes interferents and allows for extensive analysis of the enzyme's properties. The extraction and purification of olive PPO is inherently a challenge due to the presence of coextracted interferents (i.e., olive oil, pectins, and pigments). This work represents a significant advancement by developing an optimized protocol that effectively eliminates these major contaminants. The focus of the third study elucidates the effects of the protease inhibitors in suppressing enzymatic browning in potatoes. The results from this study suggests that instead of PPO, the lowered concentration of free amino acids and tyrosine, regulated through cysteine protease inhibition, is key to enzymatic browning in potatoes. This review will start by comparing and contrasting the first two papers, both of which study PPOs, albeit from different species. A discussion on the third work on protease inhibitor will then be presented
Advancements in Recommender Systems through Collaborative Filtering Techniques
No full textIn the era of digital personalization, recommendation systems are increasingly important in improving user experience as they suggest relevant content depending on the past user interactions. While addressing its inherent difficulties, including data sparsity, cold-start problems, and lack of interpretability, this thesis explores collaborative filtering (CF) as an underlying approach. Conventional memory-based and model-based CF techniques are analyzed and applied. Also, the deep learning-based recommender system is implemented using user and item embeddings combined with a multi-layer perceptron (MLP). SHAP (SHapley Additive exPlanations) is included to evaluate the model's predictions in order to improve the trust and user confidence, providing insights for both end-users and stakeholders. Through extensive experiments, various techniques reveal advantages and disadvantages among scalability, personalization, and interpretation. Overall, the evaluation supports the design of more efficient and reliable recommendation systems based on the particular data features and application needs
Field Extensions and Multiplication on their Induced Tori
No full textWe explore the relationships between quadratic fields and their respective rings of algebraic integers, focusing on two-dimensional rational vector spaces and two dimensional integer lattices. We investigate how these structures interact and examine their underlying algebraic and geometric properties. We delve into the ways in which the properties of quadratic fields influence the structure of two dimensional rational vector spaces and how their rings of algebraic integers influence two dimensional integer lattices. We also introduce a concept we call induced multiplication, which arises from vector space and Z-module isomorphisms. These isomorphisms link these distinct mathematical structures together. We leverage these relationships to construct lattices and tori which allows for geometric interpretations of quadratic fields and their integer rings. This provides valuable insight into their topological and algebraic properties, enhancing our understanding of the connection between algebra and geometry. We provide an investigation of the rational circle group, rational lattices and rational tori. Additionally, we study the endomorphisms of rational tori along with matrix representations for algebraic integers
Mutual Mate Choice and the Dynamics of a Courtship Duet in a Neotropical Frog
No full textMating systems of sexually reproductive taxa are commonly characterized by competitive males and choosy females. However, this narrow view overlooks many behaviors that shape mating systems, including dynamic interactions wherein both sexes actively court the other (in some instances, in a coordinated sequence) and mutual mate choice. At the population level, these features could have multiplicative effects on reproductive success, and, at the species level, they could facilitate the divergence of mating behaviors between geographically isolated populations. Here, we investigate courtship dynamics in an anuran mating system and test for mutual mate choice. This is one of few empirical studies on complex courtship dynamics in an anuran, and the only study, to our knowledge, of these dynamics in a frog system that lacks parental care. Our goal is to broaden our collective perspective on the evolution of anuran mating systems. Using behavioral assaying and computational approaches on a population of red-eyed treefrogs, Agalychnis callidryas, this project reveals that this mating system has 1) an interactive and low stereotypy courtship dynamic in which multiple pathways lead to reproductive success, and 2) males and females court and reject mates, indicating mutual mate choice. This study provides new insights into the mating system dynamics of Agalychnis callidryas, revealing that anuran mating systems may be more complex than previously thought. Future research should explore the mechanisms underlying anuran behaviors to expose more nuanced mating dynamics and understand alternative factors that could influence mating system reproductive success
Real-Time Renewable Energy Forecasting Using Digital Twin Models for Smart Grid Efficiency
No full textThe increasing use of renewable energy sources, like wind and solar, has created new problems for power grid management because of their unpredictability and sporadic nature. This project investigates how to enable precise, real-time forecasting in smart grid systems by combining cutting-edge machine learning models with digital twin technology. Several forecasting algorithms, such as Random Forest, Gradient Boosting, Support Vector Regression, and Long Short-Term Memory (LSTM), were applied and assessed using a comprehensive dataset that included energy metrics, environmental factors, and system behaviors. The study revealed that models like LSTM and Gradient Boosting delivered high accuracy with Mean Absolute Percentage Error (MAPE) values under 3%, making them highly suitable for practical deployment in energy prediction and smart grid optimization tasks. The combination of machine learning and digital twins improved predictive performance and made it possible for grid operations to use adaptive planning, dynamic control, and intelligent monitoring. By simulating and responding to real-time data inputs, digital twins can help utilities anticipate disruptions, reduce maintenance costs, and align energy generation with demand. This work demonstrates the transformative potential of digital twin-enabled systems for building resilient and efficient energy infrastructure. Future studies will concentrate on developing hybrid models, edge computing solutions, real-time implementation, and cybersecurity to enhance and expand smart grid applications
One dimensional model for the analysis of GaAs MESFET frequency behavior
No full textThis thesis presents a detailed analysis of the electrical performance and characteristics of Gallium Arsenide Metal-Semiconductor Field Effect Transistors (GaAs MESFETs). The study focuses on evaluating key device parameters such as drain-to-source current (IDS), drain-to-source voltage (VDS), gate-to-source capacitance (Cgs), gate-to-source voltage (Vgs), cut-off frequency (fₜ), and gate length (Lg). The IDS vs. VDS analysis indicated clear linear behavior at lower gate voltages (0.1-0.2 V) and a distinct saturation region at higher gate voltages (0.3 V), demonstrating stable transistor performance with maximum and minimum drain currents observed at approximately 2.5 mA and 1.5 mA, respectively. The analysis of the cut-off frequency versus gate length clearly highlighted the performance advantages associated with shorter gate lengths, showing a maximum cut-off frequency of approximately 2.8 × 10⁷ Hz at a gate length of 0.5 µm. This underscores the importance of precise fabrication techniques to manage short-channel effects while optimizing device performance. Additionally, the capacitance characteristics of the device were extensively studied, revealing an exponential increase in gate-to-source capacitance (Cgs) from 2.15 × 10⁻¹³ F at negative voltages (around -0.8 V) to approximately 5.5 × 10⁻¹³ F at near-zero voltages (around 0.2 V). This significant capacitance variation highlights the sensitivity of MESFET devices to gate voltage adjustments, essential for applications in high-speed and optoelectronic circuits. Collectively, these findings provide critical insights into optimizing GaAs MESFET devices, emphasizing precise control of gate dimensions, operating voltages, and diffusion charges to achieve enhanced high-frequency performance, stability, and efficiency suitable for advanced electronic and optoelectronic applications
Energy Prediction for Automobile Air Conditioning Systems through Deep Learning and Data Science
No full textThe use of automobile air conditioning systems increases gas emissions and fuel consumption, emphasizing a need for improved energy prediction in vehicles. This study investigates the energy consumption and prediction of air conditioning systems in hybrid and internal combustion engine vehicles through multiple machine learning algorithms. An OBD-II sensor was utilized to collect dynamic and static data from the vehicles along with data integrated from the OpenWeatherMap API. However, this data lacks the air conditioning-related parameter of Aircon Consumption Power. Instead, the collected data utilizes fuel rate as the target variable for modeling. To address this problem, a two-stage method was used to collect data for each trip, capturing vehicle data metrics including fuel consumption with the A/C on and off. As a result, this created a comparative approach for the analysis of fuel rates and a deeper understanding of the impact of air conditioning on energy consumption. Next, data preprocessing and cleaning were done to select relevant and highly correlated features. Given the lack of an exact air conditioning-related feature, a two-step approach took place. From there, a classification model determined whether the air conditioning system was active, and a regression model then predicted overall energy consumption based on the target variable. Once the data was collected, multiple machine learning models, including linear regression, XGBoost, and Stacking ensemble model (using Decision Tree, Random Forest, XGBoost, Light GBM, and Linear models) were used to predict energy consumption. Known for handling complex datasets, XGBoost was combined with the ensembled model to predict instantaneous fuel rates with high accuracy. Overall, the study's findings highlight the potential use of machine learning to improve energy predictions in vehicles by providing observations that could contribute to more efficient air conditioning systems
Speech recognition through smartphone gyroscope data analysis
Get PDFThis study explores the innovative use of motion sensors built into smartphones for speech recognition. Unlike traditional systems that rely on microphones, this approach employs gyroscopes and accelerometers to capture the angular velocity and orientation of the user's lip and jaw movements during silent speech. By processing this motion data through advanced machine learning techniques, including Dynamic Time Warping (DTW), Hidden Markov Models (HMM), and Neural Networks (NN), the system effectively identified spoken vowels and numbers, achieving approximately 80% accuracy. The research highlights the advantages of this method in noise-resistant environments, where microphones often struggle, and its energy efficient nature, with motion sensors consuming significantly less power compared to microphones. Experimental results demonstrated successful detection of vowels and numbers one through five, showing promise for silent speech applications and assistive technologies. This work represents a step forward in human-computer interaction, leveraging motion sensors to provide a privacy-conscious and low-power alternative to traditional audio-based speech recognition. Future enhancements could refine accuracy and enable the detection of more complex vocabulary constructed from vowels, opening new possibilities for wearable devices, mobile technology, and assistive communication systems