Aishwarya Pawar
I have 4 years of work experience as a Data Scientist. My area of expertise include using machine learning algorithms to design marketing strategies; ensuring e-commerce business conversion improvement in A/B tests, and customer behavior analysis for business strategies. I was quickly promoted to become a Scrum Master of 15-member team. However, it's the excitement of digging deeper into the numbers to find a pattern, which keeps me motivated. To fuel my passion I am currently pursuing masters in Business Analytics with machine learning major at University of Texas, Austin.
When I am not playing with data, I love to travel, and learn about new culture by meeting new people. Also, I am an avid non-fiction books reader and enjoy occasional cartoon doodling.
Experience
Zelis Healthcare
Enhancing market differentiating capability to increase the market share by employing machine learning algorithms. This algorithm will identify high value providers as well as those providers that are most likely to accept/register with Zelis for one of its electronic payment services.
Unbxd
Ensured 10+ e-commerce businesses improve their conversion using Unbxd machine learning algorithms. Worked with data science team to improve recommendation and search algorithms for product discovery on sites. Helped sales and marketing team take data-driven decisions Conducted training sessions for the team on statistics and data visualization. Helped in the recruitment processes by conducting data science interviews.
Mu Sigma
Helped the marketing team of one of the biggest retailer design multiple campaigns, and measured their successes. Analyzed customer behavior to recommend marketing and supply chain strategies using machine learning. I was awarded for my curiosity, quickly learning new technology, and thought leadership.
With fast promotion to a Scrum Master, I managed the crucial client communications of a 15-member team. I also initiated sprint ceremonies for smooth functioning of the team, and educated the team with LeanKit (agile tool). Apart from the work, I trained a SQL team of 35 new joiners and mentored a business case study for 5 new joiners.
Education
University of Texas, Austin
Coursework Includes: Marketing analytics, Supply chain analytics, Pricing and Revenue management, Financial management, Database management, Decision analysis, Advanced predictive modelling (using scikit & Tensorflow)
Pune Univeristy, India
First Class with Distinction | Dept Rank:7th
Skills
- Regression (Linear, Ridge, Lasso)
- Classification (Logistic regression, Naïve Bayes)
- Clustering (K-means, hierarchical)
- Decision Trees and Random Forest
- Bagging, Boosting
- Natual Language Processing - Sentiment Analysis, Topic Modeling
- Deep Learning with Convolutional Neural Networks (CNN- VGG16, ResNet, MobileNet)
- Design of Experiment (A/B Testing)
Projects
Corporate Projects
-
E-Commerce site A/B Testing
Collaborated with 10+ e-commerce clients (fashion, automobile, pharma, home & living etc. verticals) to ensure conversion improvement in A/B tests using statistical analysis. Acquired $100k monthly recurring revenue and the framework was adopted at the enterprise level.
-
Conversion Driver Analysis
Improved conversion of a market place e-commerce site by 11% using driver analysis. Also, educated the client with their analytics platform, and this helped the company build strong relationship with the client. Additionally, this analysis became a mandatory step in customer on-boarding process.
Analytics Tools: Random Forest, R, SQL (BigQuery)
-
Customer acquisition analysis
Acquired 45% of the targeted customers (in 2 weeks) by identifying potential grocery service adopters for personalized marketing campaign.
Analytics Tools: Logistic Regression, R, SQL (Hive)
-
Customer Focus Group Identification, Profiling, and opportunity sizing
Defined a customer focus group, worth $40B, based on their time and money constraints using the data. As this was the most important customer group for the year, we built dashboards using tableau and D3 (JavaScript) to track their changing behavior. I was honoured with Star Performer Award (quarterly to top 3% achievers ) for perspicacity, and thought leadership.
-
Designed an experiment to understand customer's perception of in-store music
Working on this challenging open ended problem, I designed a fractional factorial design of experiment using transactional and demographic data of the stores. I clustered the stores to have replication and blocking of the experiment. The client highly appreciated the roadmap in the business review.
Analytics Tools: Regression driver analysis, K-mediodes clustering, fractional factorial design of experiment tool
-
Holiday Server traffic forecasting
Forecasted server traffic for holiday to prevent downtime, which has costed $80K in the previous year. The forecast was with 4% offset from the actual traffic, which ensured 99% SLA as promised to the customers.
-
Business Review Report Automation and executive dashboard
Automated the manual reports which took ~120 person-hours a year. Created a dashboard with real-time data. Additionally, I set up alerts for alarming peaks or drops to the engineering team to debug for any potential platform failure. Executives loved this dashboard and decided to host it on the office LCD 24x7 for a sanity check. The dashboard also became a part of sales pitch to the new customers.
Analytics Tools: Google App Script, SQL(Big Query), Data Studio, Unbxd API
E-Commerce site A/B Testing
Collaborated with 10+ e-commerce clients (fashion, automobile, pharma, home & living etc. verticals) to ensure conversion improvement in A/B tests using statistical analysis. Acquired $100k monthly recurring revenue and the framework was adopted at the enterprise level.
Conversion Driver Analysis
Improved conversion of a market place e-commerce site by 11% using driver analysis. Also, educated the client with their analytics platform, and this helped the company build strong relationship with the client. Additionally, this analysis became a mandatory step in customer on-boarding process.
Analytics Tools: Random Forest, R, SQL (BigQuery)
Customer acquisition analysis
Acquired 45% of the targeted customers (in 2 weeks) by identifying potential grocery service adopters for personalized marketing campaign.
Analytics Tools: Logistic Regression, R, SQL (Hive)
Customer Focus Group Identification, Profiling, and opportunity sizing
Defined a customer focus group, worth $40B, based on their time and money constraints using the data. As this was the most important customer group for the year, we built dashboards using tableau and D3 (JavaScript) to track their changing behavior. I was honoured with Star Performer Award (quarterly to top 3% achievers ) for perspicacity, and thought leadership.
Designed an experiment to understand customer's perception of in-store music
Working on this challenging open ended problem, I designed a fractional factorial design of experiment using transactional and demographic data of the stores. I clustered the stores to have replication and blocking of the experiment. The client highly appreciated the roadmap in the business review.
Analytics Tools: Regression driver analysis, K-mediodes clustering, fractional factorial design of experiment tool
Holiday Server traffic forecasting
Forecasted server traffic for holiday to prevent downtime, which has costed $80K in the previous year. The forecast was with 4% offset from the actual traffic, which ensured 99% SLA as promised to the customers.
Business Review Report Automation and executive dashboard
Automated the manual reports which took ~120 person-hours a year. Created a dashboard with real-time data. Additionally, I set up alerts for alarming peaks or drops to the engineering team to debug for any potential platform failure. Executives loved this dashboard and decided to host it on the office LCD 24x7 for a sanity check. The dashboard also became a part of sales pitch to the new customers.
Analytics Tools: Google App Script, SQL(Big Query), Data Studio, Unbxd API
Academic Projects
-
Deep Learning : Multi-Label Image Classification
With recent the Amazon rainforest fire and Australia bushfire, there is a serious need to understand the severity of deforestation problem. In this project, I identified the reasons for deforestation of the Amazon forest using multi-label classification image classification of it's satellite images. Additionally, I solved the problem of class-imbalance using unconventional ensemble learning to increase the recall.
Analytics Tools: Convolutional Neural Networks (10-layer CNN, VGG-16, ResNet, MobileNet), XGBoost, Python, Google Cloud Platform
-
Marketing Mix Modeling
"Half the money I spend on advertising is wasted; the trouble is, I don't know which half." - John Wanamaker E. Jerome. To understand which half of ad spend is working, I analyzed 4 Ps (Product, Promotion, Price, and Place) marketing mix for automobile industry. Furthermore, I used Dorfman-Steiner rule to understand if different product types have the optimal mix of price and ad spend.
Analytics Tools: Linear Regression (OLS, Log-log), Python
-
Natural Language Processing
-
CEO Tweets analysis
With increasing marketing strategies on social media platforms, we have witnessed how Elon Musk's tweet about Tesla going private impacted the stock market. Do the CEO tweets have influence on the stock market? Using the machine learning, I found out the answer to be 'Yes'. So, what attributes make the tweet influential? Can a CEO be more influential by including/avoiding some factors in her tweets? In this project, I found out the marketing strategies for CEOs to increase their influence on how they could attempt to influence the stock markets.
Analytics Tools: Linear Regression, Logistic Regression, Sentiment Analysis, Topic Modeling (LDA), Python
-
Is a picture worth thousand words on Instagram?
Growing up we all remember watching the engaging National Geography programs. With growing competition, National Geography wants to understand customer preference using their posts on Instagram. In this project, I analyzed influencing posts and what are some common topics people are interested about.
Analytics Tools: Logistic Regression, Topic Modeling (LDA), Python, Google Image Extraction API
-
Automobile Discussion Forum Analysis
Brand perception is very important in the automobile industry. In this project, I analyzed an automobile discussion forum to understand what attributes top brands are associated with and which other brands they are compared to. Additionally, I suggested some business strategies for brand and marketing managers to improve their brand perception for their target customer groups.
Analytics Tools: Association Analysis using lifts and MDS plots, Python, Google Image Extraction API
-
Bank Customr Churn Analysis
Customer churn is a major problem for any organization. It becomes important to identify these customers for business strategies with limited budget. In this project, I experimented with multiple algorithms to obtain the maximum accuracy for prediction of churned customers.
Analytics Tools: Logistic Regression, KNN, Decision Tree, Random Forest, Python
-
NBA Players Salary Prediction
Ever wondered about how much your favorite NBA player's worth the next season? Experimenting with different machine learning algorithms, here is our attempt to accurately predict NBA player's salaries. We also diagnosed the limitation of these algorithms.
Machine Learning Algorithms: Linear Regression (Ridge, Lasso), Random Forest, Bagging, Boosting, and Neural Network, R
Deep Learning : Multi-Label Image Classification
With recent the Amazon rainforest fire and Australia bushfire, there is a serious need to understand the severity of deforestation problem. In this project, I identified the reasons for deforestation of the Amazon forest using multi-label classification image classification of it's satellite images. Additionally, I solved the problem of class-imbalance using unconventional ensemble learning to increase the recall.
Analytics Tools: Convolutional Neural Networks (10-layer CNN, VGG-16, ResNet, MobileNet), XGBoost, Python, Google Cloud Platform
Marketing Mix Modeling
"Half the money I spend on advertising is wasted; the trouble is, I don't know which half." - John Wanamaker E. Jerome. To understand which half of ad spend is working, I analyzed 4 Ps (Product, Promotion, Price, and Place) marketing mix for automobile industry. Furthermore, I used Dorfman-Steiner rule to understand if different product types have the optimal mix of price and ad spend.
Analytics Tools: Linear Regression (OLS, Log-log), Python
Natural Language Processing
-
CEO Tweets analysis
-
Is a picture worth thousand words on Instagram?
-
Automobile Discussion Forum Analysis
With increasing marketing strategies on social media platforms, we have witnessed how Elon Musk's tweet about Tesla going private impacted the stock market. Do the CEO tweets have influence on the stock market? Using the machine learning, I found out the answer to be 'Yes'. So, what attributes make the tweet influential? Can a CEO be more influential by including/avoiding some factors in her tweets? In this project, I found out the marketing strategies for CEOs to increase their influence on how they could attempt to influence the stock markets.
Analytics Tools: Linear Regression, Logistic Regression, Sentiment Analysis, Topic Modeling (LDA), Python
Growing up we all remember watching the engaging National Geography programs. With growing competition, National Geography wants to understand customer preference using their posts on Instagram. In this project, I analyzed influencing posts and what are some common topics people are interested about.
Analytics Tools: Logistic Regression, Topic Modeling (LDA), Python, Google Image Extraction API
Brand perception is very important in the automobile industry. In this project, I analyzed an automobile discussion forum to understand what attributes top brands are associated with and which other brands they are compared to. Additionally, I suggested some business strategies for brand and marketing managers to improve their brand perception for their target customer groups.
Analytics Tools: Association Analysis using lifts and MDS plots, Python, Google Image Extraction API
Bank Customr Churn Analysis
Customer churn is a major problem for any organization. It becomes important to identify these customers for business strategies with limited budget. In this project, I experimented with multiple algorithms to obtain the maximum accuracy for prediction of churned customers.
Analytics Tools: Logistic Regression, KNN, Decision Tree, Random Forest, Python
NBA Players Salary Prediction
Ever wondered about how much your favorite NBA player's worth the next season? Experimenting with different machine learning algorithms, here is our attempt to accurately predict NBA player's salaries. We also diagnosed the limitation of these algorithms.
Machine Learning Algorithms: Linear Regression (Ridge, Lasso), Random Forest, Bagging, Boosting, and Neural Network, R
Awards & Accomplishments
- Master of Science in Business Analytics from Univeristy of Texas, Austin (2020)
- Star Performer award for quickly learning new technology, and ownership for quality deliverables (2016)
- SQL Mentor for 35 new joiners (2016) and business case study mentor for 5 new joiners (2017) at Mu Sigma University
- Merit for being among the top 2% of the Pune University's Electronics and Telecommunication engineering department (2015)
- Qualified DELE A2 Spanish exam conducted by Government of Spain (2014)
- Awarded National English merit for being among the top 0.1% of the successful candidates by CBSE (2009)
- Awarded regional level government merit scholarship for mathematics and verbal abilities (2004 - 2007)