Harsh Agarwal

Highly skilled data analyst with a master's degree in data science and a proven track record of leveraging analytical techniques to extract actionable insights from complex datasets. Committed to driving data-driven decision-making and delivering valuable solutions to optimize business performance. Seeking a challenging position in a dynamic organization where I can apply my technical expertise, project experience, and strong problem-solving abilities to contribute to strategic initiatives and help achieve organizational goals. Dedicated to utilizing advanced statistical modeling, machine learning algorithms, and data visualization techniques to unlock the full potential of data and empower stakeholders with actionable recommendations for improved operational efficiency, revenue growth, and customer satisfaction.

1. Mushroom Classification and Analysis:

• Conducted a comprehensive analysis of the "Mushroom.csv" dataset, applying data cleaning and exploratory data analysis techniques to understand the attributes and relationships within the dataset.
• Developed and implemented classification models, including decision trees, random forests, and logistic regression, to accurately predict the edibility of mushrooms, achieving [98.2%] accuracy in classification.
• Communicated project findings through effective data visualizations and documented the entire project process, methodologies, and results in a comprehensive report.

2.Deep Learning Model Development and Evaluation:

• Developed a deep learning model for ECG Heartbeat Classification, leveraging techniques such as exploratory data analysis (EDA) and hyperparameter tuning.
• Applied suitable deep learning models and preprocessed the dataset to ensure compatibility with the chosen models.
• Built, trained, and validated deep learning models, optimizing their performance through hyperparameter tuning and evaluating their effectiveness using metrics like precision, recall, F1-score, ROC-curve, and PR-curve.

3. UK Violent Crime and Firearms Analysis:

• Conducted analysis to verify claims made in the documentary regarding violent crime in the UK.
• Utilized publicly available datasets, including UK Home Office's Street Level Crime Data and English Indices of Deprivation Data.
• Employed Apache Spark on a cloud IaaS platform to efficiently process the data.
• Filtered the dataset for relevant crimes and assessed trends in violent crimes.
• Investigated the claim of higher firearms incidents per capita in Birmingham compared to other UK locations.
• Explored the association between firearms incidents and drugs offenses.
• Created informative visualizations to support the analysis
• Presented findings and analysis in a Jupyter Notebook.

4. TRAVELNORTH Database Analysis and Implementation

• Applied advanced data modeling, database design, and implementation techniques to address a complex data management problem within the TRAVELNORTH scenario.
• Utilized Oracle database system to develop data warehousing and data analytics solutions.
• Analyzed, designed, and implemented a logical relational schema, ensuring data reliability and adherence to naming conventions.
• Populated the database with sample data and performed queries using Relational Algebra and SQL.
• Explored aspects of object-relational and NoSQL database implementations for enhanced functionality.
• Produced a comprehensive report addressing professional, legal, ethical, and security considerations for TRAVELNORTH.

5.  Data Warehousing and Data Mining

• Improved performance in data warehousing tasks by designing and implementing optimized indexes.
• Created new materialized views to enhance query performance and user experience.
• Utilized database optimization techniques to rewrite queries and leverage materialized views.
• Conducted data mining analysis on the UnitedCreditCards dataset to predict credit card default payments.
• Developed predictive models using suitable algorithms and evaluated their capabilities.
• Provided recommendations to the UNITED FINANCE company based on the findings.
• Critically evaluated data quality and adherence to standards in the SH data warehouse and UnitedCreditCards dataset.

6. Dissertation: House Price Prediction using Machine learning incorporating non-visual parameters.

• Developed a machine learning framework to predict house prices by incorporating non-visual characteristics, such as socioeconomic factors, neighborhood features, transit accessibility, and environmental considerations.
• Implemented various machine learning algorithms, including Random Forest, XGBoost, and LightGBM, to build predictive models.
• Conducted feature engineering to extract meaningful insights from the data and improve model performance.
• Evaluated model performance using metrics like RMSE, R-squared, and MAE to assess accuracy and predictive power.
• Collaborated with a team to collect and preprocess relevant datasets, perform data exploration, and validate the models.
• Presented project findings to a panel of industry experts, highlighting the significance of non-visual factors in house price prediction.