Tanya Singh

Versatile Operations Specialist successful at taking on routine and complex business challenges with resourceful and creative approach. Recognized as driven, dedicated and hardworking leader with history of streamlining operations and increasing results.

1.Machine Learning - Healthcare Provider Fraud Detection Analysis

>Chose Decision Tree model with GridsearchCV for analysis, achieving a 77% accuracy score.
>Demonstrated 80% precision in positive case predictions, indicating superior model performance.
>Achieved a strong F1 score of 76%, emphasizing a balance between precision and recall.
>True positive rate at 84%, surpassing other models and providing high inter-rater reliability (kappa score: 0.746).
>AUC score of 76% for class separability, with GridsearchCV yielding the highest score.
>Noted challenges in working with large datasets but highlighted adaptability of Decision Trees to handle various data types, without requiring normalization.
>Emphasized the scale-invariant nature of Decision Trees and Random Forests as machine learning approaches.

2. Cricket Player Performance Analysis:
 >A SQL project focused on analyzing ICC Test cricket data, extracting valuable insights and trends to inform strategic decision-making for teams and players' performance evaluation. 

>Designed and implemented complex SQL queries to retrieve and manipulate large datasets, leveraging aggregate functions, joins, subqueries, and advanced filtering techniques to uncover patterns, batting averages, bowling statistics, and other key metrics.
>Utilized SQL functions and techniques to calculate performance indicators such as strike rates, economy rates, and batting averages, enabling the identification of top-performing players and areas for improvement within the teams.

1.Machine Learning - Healthcare Provider Fraud Detection Analysis

>Chose Decision Tree model with GridsearchCV for analysis, achieving a 77% accuracy score.
>Demonstrated 80% precision in positive case predictions, indicating superior model performance.
>Achieved a strong F1 score of 76%, emphasizing a balance between precision and recall.
>True positive rate at 84%, surpassing other models and providing high inter-rater reliability (kappa score: 0.746).
>AUC score of 76% for class separability, with GridsearchCV yielding the highest score.
>Noted challenges in working with large datasets but highlighted adaptability of Decision Trees to handle various data types, without requiring normalization.
>Emphasized the scale-invariant nature of Decision Trees and Random Forests as machine learning approaches.

2. Cricket Player Performance Analysis:
 >A SQL project focused on analyzing ICC Test cricket data, extracting valuable insights and trends to inform strategic decision-making for teams and players' performance evaluation. 

>Designed and implemented complex SQL queries to retrieve and manipulate large datasets, leveraging aggregate functions, joins, subqueries, and advanced filtering techniques to uncover patterns, batting averages, bowling statistics, and other key metrics.
>Utilized SQL functions and techniques to calculate performance indicators such as strike rates, economy rates, and batting averages, enabling the identification of top-performing players and areas for improvement within the teams.