Papers by SHREYAS MAHIMKAR
TIJER, 2023
Predicting crime locations is a critical component of modern law enforcement strategies, aiming t... more Predicting crime locations is a critical component of modern law enforcement strategies, aiming to enhance public safety and optimize resource allocation. This research explores the application of big data analytics and Map-Reduce techniques to improve the accuracy of crime location predictions. As urban areas grow and crime data becomes increasingly complex, traditional methods of crime forecasting often fall short. By leveraging the power of big data and advanced analytics, this study seeks to address these limitations and offer a more robust framework for predicting criminal activity. Data for this study was collected from various sources, including police reports, public crime databases, and social media feeds. The dataset encompasses a broad range of variables such as crime type, location, time of occurrence, and demographic information. The application of Map-Reduce techniques allowed for the distribution of data processing tasks across multiple servers, significantly reducing computation time and enabling real-time analysis. The research employs several big data analytics methods, including spatial clustering, temporal analysis, and predictive modeling. By integrating Map-Reduce, the study was able to scale these methods to handle large datasets efficiently, providing more accurate and timely predictions. The results indicate that big data analytics combined with Map-Reduce techniques significantly enhance the precision of crime location predictions. The analysis revealed distinct crime patterns and trends, which can be used to inform law enforcement strategies and allocate resources more effectively. The study also highlights the benefits of real-time data processing in improving predictive accuracy and responsiveness.
IJNTI, 2023
The exponential growth of TV viewership data has necessitated the development of advanced analyti... more The exponential growth of TV viewership data has necessitated the development of advanced analytical techniques to extract actionable insights for broadcasters and advertisers. This paper explores the application of Apache Spark and Scala for analyzing large-scale TV viewership data, focusing on extracting meaningful patterns and trends that can inform strategic decisions in media planning and advertising. Apache Spark, a distributed data processing framework, is particularly well-suited for handling vast amounts of data efficiently, while Scala, as a language integrated with Spark, offers robust functional programming capabilities that enhance data processing tasks. The study begins with a detailed review of TV viewership data types and the challenges associated with managing and analyzing such data. TV viewership data typically includes metrics such as audience ratings, viewing duration, and demographic information. The paper discusses how traditional data processing methods fall short in handling the volume and complexity of this data, leading to the adoption of Spark and Scala. We then outline the methodology for leveraging Spark's in-memory processing capabilities to perform data transformations and aggregations. Using Scala, we implement data cleaning, feature extraction, and statistical analysis routines. The paper presents several case studies demonstrating how Spark and Scala can be used to uncover trends in viewership patterns, such as peak viewing times, audience preferences by genre, and the effectiveness of advertising campaigns. Key findings highlight the efficiency of Spark's distributed computing model in reducing processing times for large datasets, compared to conventional data processing tools. Scala's functional programming paradigm facilitates the development of complex data pipelines that are both scalable and maintainable. The integration
JETNR, 2024
In the rapidly evolving landscape of television advertising, accurate data modelling is crucial f... more In the rapidly evolving landscape of television advertising, accurate data modelling is crucial for understanding and optimizing advertising metrics. This paper explores data modelling techniques in both SQL and NoSQL environments, focusing on their applications and effectiveness in managing TV advertising data. Traditional SQL databases have long been the cornerstone of data management, offering robust, structured data modelling capabilities ideal for transactional systems and complex queries. In contrast, NoSQL databases, with their flexible schemas and scalability, have emerged as powerful tools for handling the diverse and voluminous data generated in the modern advertising ecosystem.
IJIRT, 2024
Effective audience targeting is crucial for optimizing TV advertising strategies and maximizing v... more Effective audience targeting is crucial for optimizing TV advertising strategies and maximizing viewer engagement. Traditional methods of audience segmentation often fall short in capturing the complex, multidimensional nature of viewer preferences. This research paper explores the use of K-Means clustering, a widely adopted unsupervised machine learning algorithm, to enhance TV viewership targeting. By employing K-Means clustering, we aim to refine audience segmentation, enabling TV networks and advertisers to better tailor their content and advertising strategies to distinct viewer segments. In this study, we utilized a comprehensive dataset comprising TV viewership data, including demographic information, viewing habits, and program preferences. The K-Means algorithm was applied to segment viewers into distinct clusters based on their viewing patterns. The process involved several key steps: data preprocessing, feature selection, and model training. Data preprocessing included handling missing values, normalizing data, and selecting relevant features to ensure the clustering results are meaningful and actionable. The K-Means clustering algorithm was configured with varying numbers of clusters to identify the optimal segmentation that provides the most insightful and actionable results. The silhouette score and elbow method were used to determine the optimal number of clusters, ensuring that the segmentation reflects meaningful distinctions between viewer groups. Our analysis revealed several distinct viewer segments, each characterized by unique viewing behaviors and preferences. For instance, one segment showed a high affinity for sports and news programming, while another group preferred drama and entertainment content. These insights allow for more targeted advertising campaigns and content recommendations, enhancing viewer engagement and satisfaction. The results of this study demonstrate that K-Means clustering provides a robust framework for improving TV viewership targeting. By segmenting viewers more
IJCSPUB, 2020
The rapid evolution of cloud computing has significantly transformed how organizations deploy and... more The rapid evolution of cloud computing has significantly transformed how organizations deploy and manage applications, with serverless platforms offering an innovative approach to software development. This paper provides a comprehensive analysis of two prominent serverless platforms: Amazon Bedrock and Claude 3. Amazon Bedrock, a part of Amazon Web Services (AWS), offers a suite of fully managed services that enable developers to build and deploy applications without the need for server management. It supports seamless integration with other AWS services, ensuring scalability, reliability, and cost efficiency. On the other hand, Claude 3, developed by Anthropic, represents a next-generation AI-driven serverless architecture that emphasizes simplicity and ease of use while leveraging artificial intelligence to optimize resource allocation and application performance. This paper compares these platforms across several dimensions, including architecture, deployment processes, scalability, cost-effectiveness, security, and ease of use. Furthermore, it explores the unique features of each platform, such as Amazon Bedrock's deep integration with AWS services and Claude 3's AI-driven optimizations. Through a series of use case scenarios, the paper highlights the advantages and limitations of each platform, providing insights into their suitability for different application requirements. By examining real-world applications and performance benchmarks, this paper aims to guide organizations in selecting the most appropriate serverless platform for their needs, considering factors such as application complexity, development speed, and operational cost. The analysis concludes with recommendations for organizations looking to leverage serverless architectures to enhance their operational efficiency and scalability.
TIJER, 2020
Predicting TV audience ratings is crucial for broadcasters and advertisers aiming to optimize con... more Predicting TV audience ratings is crucial for broadcasters and advertisers aiming to optimize content scheduling and maximize advertising revenue. Traditional methods of audience rating prediction often lack precision due to the complexity of viewer behavior and the dynamic nature of television consumption. This research explores the application of linear regression models to enhance the accuracy of TV audience rating predictions. Linear regression, a foundational statistical technique, models the relationship between a dependent variable and one or more independent variables, offering a straightforward approach to forecasting ratings.
IJRAR, 2021
Predictive analysis of TV program viewership has become increasingly crucial in the media industr... more Predictive analysis of TV program viewership has become increasingly crucial in the media industry as broadcasters and advertisers seek to optimize content delivery and maximize engagement. This study explores the application of Random Forest algorithms to predict TV program viewership trends, aiming to enhance forecasting accuracy and inform strategic decision-making. Random Forest, a versatile ensemble learning technique, combines multiple decision trees to improve predictive performance and handle complex datasets. The research begins with an overview of the significance of predictive analytics in television programming. Accurate predictions of viewership patterns allow media companies to tailor content, optimize scheduling, and make informed advertising decisions. The study emphasizes the need for advanced analytical methods to navigate the voluminous and heterogeneous nature of TV viewership data. Data for the analysis was sourced from multiple channels, including historical viewership records, demographic information, and social media engagement metrics. The dataset was preprocessed to handle missing values, normalize features, and encode categorical variables. Random Forest algorithms were then employed to model the relationships between various predictors, such as time slots, program genres, and viewer demographics, and the target variable of viewership numbers.
JETIR, 2021
The effectiveness of TV advertising campaigns has become increasingly critical in the highly comp... more The effectiveness of TV advertising campaigns has become increasingly critical in the highly competitive media landscape, necessitating advanced methods to accurately measure and optimize advertising impact. This study explores the application of lift and attribution models to evaluate TV advertising campaign effectiveness, with a particular focus on integrating survey data for a comprehensive analysis. Lift models, which measure the incremental effect of advertising on consumer behaviour, and attribution models, which allocate credit to different touchpoints in the customer journey, are examined for their effectiveness in assessing advertising impact.
IJNRD, 2022
In the rapidly evolving field of consumer electronics, ensuring high data quality is paramount fo... more In the rapidly evolving field of consumer electronics, ensuring high data quality is paramount for driving innovation, enhancing user experiences, and maintaining competitive advantage. This paper explores the application of big data techniques to address and improve data quality issues within the consumer electronics industry. As the volume, variety, and velocity of data generated by modern electronic devices increase, so too do the challenges associated with maintaining accurate, reliable, and usable data. This research aims to investigate how big data methodologies can be leveraged to enhance data quality and proposes strategies for implementing these techniques effectively. The study begins with a comprehensive review of existing literature on data quality challenges specific to consumer electronics By integrating big data techniques such as data cleaning, validation, and advanced analytics, the research aims to provide solutions to these challenges. To empirically validate the effectiveness of these techniques, the research incorporates survey data from industry stakeholders, including data engineers, product managers, and consumers. The survey explores their perceptions of current data quality practices, the impact of data quality issues on their operations, and their experiences with big data solutions. The survey results offer valuable insights into the practical implications of data quality management and the perceived benefits and limitations of big data techniques. Key findings indicate that big data techniques, including machine learning algorithms, data warehousing, and real-time analytics, significantly enhance data quality by automating data validation processes, detecting anomalies, and providing actionable insights. The research concludes with recommendations for consumer electronics companies to adopt a holistic approach to data quality management, combining technological solutions with robust data governance practices. Emphasis is placed on the importance of ongoing training for staff, investment in advanced analytics tools, and the development of comprehensive data quality frameworks.
IJCRT, 2022
The dynamic landscape of television viewership is influenced by numerous factors, including demog... more The dynamic landscape of television viewership is influenced by numerous factors, including demographic shifts, content preferences, and viewing habits. To navigate these complexities, media companies and advertisers increasingly turn to predictive modelling techniques powered by M.L. This study explores the application of ML in forecasting TV viewership trends, aiming to enhance understanding and anticipation of audience behaviour.
Uploads
Papers by SHREYAS MAHIMKAR