I (Yashaswini Dhatrika) am a second-year Master Student in Data Science at Indiana University Bloomington, with a solid data science background, ~1 year of research experience, and ~3-years of professional experience in Machine Learning and Statistical Modelling.
During my professional experience, I have worked on end to end on projects that are right from the inception of an idea, converting to analytics problem then till generating insights and convincing the client about the insights generated. My research experience has been experimenting with a new machine-learning algorithm and tweaking the existing one. During this time, I have realized that in this era of data, data science is a great tool to create impacts. So, my inclination is to create an impact in different businesses using the powerful tools data science and machine learning.
-
Experience
Data Analyst | Kelley School of Business, USA (June 2019 – Present)
• Developed a couple of workflows to analyze the performance of the various campaign in Alteryx that inputs the data of various sources from Google Analytics, Vendor Database, and Salesforce and then pushes the cleansed and transformed data to the MS SQL Database.
• Performed channel attribution Markov model to identify the effective channels and paths contributing to conversions.
Revenue & Planning Analyst | IndiGo Airlines, India (May 2017 – July 2018)
• Developed a forecasting engine using LSTM to forecast the demand of the passengers for ~1000 flights which operated ~70 destinations. The optimized engine showed the Revenue (Passenger revenue per available seat kilometer) a 7.7% increase for 6 months compared to the prior year for the same period.
Business Analyst | BRIDGEi2i Analytics Solutions Private Limited, India (Jun 2015 – May 2017)
Awards: Awarded the Best Team award twice for a period of 2 years for providing the quality & impactful insights for the client.
• Developed a propensity model (Logistic Regression) to prioritize the customers and identify effective means of communication for cross-selling across multiple products for a financial company. The impact of the implementation showed improvement in the cross-sell conversion by 5% compared to prior months
• Analyzed online purchase intentions and developed a propensity model (Logistic Regression, Spark) using clickstream data for a technology firm to identify the best prospects for targeted marketing efforts. The impact of the implementation showed improvement in the conversion by 13% compared to prior months.
-
Projects
Customer Profiling & Lead Prioritization Modeling:
Tools/Concepts: R, SQL, Logistic Regression.
• Built scorecard models using logistic regression to prioritize leads and efficiently allocate resources for marketing campaigns.
• Helped the client in reducing the acquisition cost of new customers up to 11% of the prior year using purchased data.
Classification algorithm to determine in-risk flights:
Tools/Concepts: R, Hive, Random forest and SVM
• Identified risk flights using random forest based on current bookings, remaining seats, booking velocity, etc. The algorithm helped the flight analyst to put the relevant fare points in the reservation system and helped in boosting the PRASK (Passenger revenue per available seat kilometer) by 3.7%.
Implementation of Text to SQL for NLIDB systems:
Tools/Concepts: Python, NLP, Probabilistic Model (Hidden Markov Models, Maximum Entropy Markov models and Conditional Random fields) & Deep Learning Models (Encoder-Decoder Architecture and the Attention Based architecture)
• Developed a system based on a probabilistic and deep learning approach that can yield SQL queries automatically by simply interpreting the natural language queries.
• Achieved accuracy of ~80% by deploying the deep learning model