Hypothesis Testing and Statistical Inference on Nationality of Artists
Continuing from visualizing the dataset of MoMA’s art collection, I am using the bootstrap method to make inferences about the population in cases wherein there is only a limited amount of samples or subset of data from the larger population, given a chosen population parameter.
Moreover, with a hypothesis test, I can compute the likelihood of my test statistics approximating the value of the real population parameter well or not. A well-designed hypothesis test includes a null and alternative hypothesis and p-value or z-score or confidence interval to describe how well this statistical model compares sample data against the population data.
I selected the dataset of artists and artworks from the Museum of Modern Art in New York City as part of my ongoing research focus on art. With over 139,000+ works by 15,000+ artists spanning across 170 years, the dataset is rich with attributes of 29 columns and updated monthly.
Its cultural significance as one of the foremost art collections worldwide could shed light on where the art world once was and where it is trending towards now. I examined my areas of interest in gender, nationality, diversity of categories of art, age of artists and artworks, and sources of collections.
By storytelling with graphs, charts, and lines, I was able to identify patterns, extract insights and visualize similarities or discrepancies that might not be self-evident when working with these large datasets. Through the series of data visualizations on 2.18 billions of data points, I am hoping to inform how we could foster a more open, diverse, and enriching tomorrow for the art world at large.
data processing, data wrangling, data transformation, data visualization, panda, python, dataframe, NumPy, matplotlib, seaborn
A sentiment analysis and data visualization of Twitter feeds and popular emoticons on the world’s top travel destinations
What emotions do different places evoke? We built an information organization and retrieval system to allow users to see the real sentiment and emotions associated with each travel destination. We rely on unaltered twitter feeds and tweets for some of the world’s most famous tourist destinations based on their geotagged location on a week’s worth of data.
We hypothesize emotion could be inferred from text-based sentiment and the popular usage of emoticons on Twitter. To help our users filter locations by emoticons and filter sentiment by locations, we used sentiment analysis from VADER library to classify all tweets, either geotagged or keyword-based. Classifying tweets by sentiment at each destination or each emoticon, combined with the frequency count by each emoticon, became the central focus of our organization system.
Moreover, to enable fast, lightweight, and intuitive queries with data, we created a menu-based system of interaction design where the user is encouraged to explore most commonly asked queries and access clear, geospatial maps or bar visualizations, and deep dive into data in a more accessible, visual format.
data clean-up, data engineering, classification, sentiment analysis, data visualization, user interaction
GoCart is your friendly, neighborhood grocery cart that acts as a shopping companion. In a world with so many choices, choosing the right foods for you is a constant challenge ‒ especially since everyone isn’t the same.
First, GoCart helps personalize your grocery runs through a quick and easy survey. However, GoCart doesn’t stop there. At the grocery store, GoCart helps encourage good behavior and discourage bad behavior with haptic feedback through shakes and vibrations and audio cues. It is personality generated according to your preference, as well a customized look.
Beyond a shopping cart, GoCart is no longer a tool in the grocery, it is your buddy to help you become a better person.
user experience research, design, prototype, user interaction mapping, user journey
Consulted as product lead to strengthen proprietary A/B testing tool for launching 200+ experiments annually on 55MM+ users worldwide
• Gathered requirements from engineering, data science, and design to build a product roadmap with focus on statistical prediction modeling, custom analytics, and usability
• Spec-ed features, wrote user-stories, and ran daily stand-ups for team of four
• Launched personalization and recommendation algorithms to drive acquisition
ARTHOUR connects travelers to local art and good guides for tours. ARTHOUR enables a different way to experience local culture and arts, and connect deeper to the insiders' community more meaningfully beyond selfies or instagram posts at Times Square. Our mission is to ensure travelers feel like they are exploring with a local friend, a smart expert, and a warm host all-in-one. Launched MVP. In user testing.
ideation, market analysis, design, prototyping, market validation, user testing, entrepreneurship, coding, infrastructure, development operations, launch, product marketing
After conducting a series of customer interviews, it became clear Saks Off Fifth's .com's checkout process was unnecessarily difficult for all customers to make a purchase. Increasingly, these customers are shopping via their mobile phones and demanding greater simplicity, clearer navigation, and faster technical delivery with no tolerance for any hiccups. Now that more than half of the customers are buying on mobile and contributing more mobile spend to the overall online revenue, this undertaking sets in stone how the business could capture and convert more digital customers in a whole new way.
• Led a cross-functional team of ten to redesign and launch responsive web checkout resulting in improved conversion rate by 6% and incremental $7 MM+ revenue
• Synthesized requirements from stakeholders and sources of customer feedback to distill them into actionable product roadmaps and evangelize across large matrixed company
conversion optimization, daily scrum, usability research, ab testing, data analytics, stakeholder management
Led cross-functional, distributed team of 10 for mobile ecosystem: mobile website, iOS, and Android apps
• Drove 180%+ revenue growth by managing product backlogs and feature priority
• Measured adoption, usage data, and business impact of bi-weekly launches
• Iterated conti¬¬nuously to optimize conversion funnels, lift KPIs, and delight customers from customer feedback, competitive analysis, and organization strategy
product strategy, roadmap prioritization, cross-functional production, daily scrum, UI/UX design, ab testing, data analytics, conversion optimization, marketing metrics
Served as day-to-day account and project manager for cross--functional team of 13 for Aetna
• Scoped four projects, each $1-3MM, to plan 20+ sprints with system interdependencies, technical architecture, UI/UX, resourcing, scheduling, risk and change negotiations
• Built three global mobile marketing strategies for General Electric, Adidas, and Clinique
• Achieved 15% lift in app engagement rate at 6 BB reach by optimizing monthly campaigns, identifying quarterly key success metrics, and monitoring analytics for mobile acquisition channels: search, display, SMS, and app store optimization (ASO)