harshvardhaniimi/map-of-tiny-perfect-things

A Map of Tiny Perfect Things

Hi there, I'm Harshvardhan! Welcome to my Github. 👋

I'm passionate about using analytical techniques to solve real-world problems. When I'm not deep in data, you'll find me engrossed in a book📚, out on a hike🥾, experimenting in the kitchen🥘, or unwinding with music🪕.

Links

• Personal Website: https://www.harsh17.in/
• Blog: https://blog.harsh17.in/
• LinkedIn: https://www.linkedin.com/in/harshvardhaniimi/
• Spotify: https://open.spotify.com/user/6z25jqql0glx8o5aqhp8xkumr
• Threads: https://www.threads.net/@harsh17.in

Present

I'm currently a doctoral student of Business Analytics and Statistics at the Haslam College of Business, University of Tennessee. My research area is business analytics, where I identify interesting applications of machine learning and statistics to elevate business value. I was also an intern with HP Inc.'s Strategic Planning and Modelling Team, where I built ML forecasting models to forecast the demand for printers worldwide. Previously, I completed my BA and MBA from IIM Indore in April 2021, and being an ERASMUS+ exchange student with EU scholarship at the University of Latvia from Feb-Jun 2019.

I've also worked on many projects from forecasting demands of 10,000+ products in 40+ geographies at HP Inc., Kaya, my GPT-powered pet cat, R packages on Indian postal codes, vaccine demand predictions, and marketing of humanoid robots.

Writing

I share my learning journey and more in a free monthly newsletter - Next: Today I Learned About Data Science. It's a curated collection of Data-Sciency works and has found resonance with 300+ subscribers. Other than that, I often write on my Blog.

Skills and Interests

I'm always open to collaborating on interesting projects, especially those that challenge me to learn something new. Here's what I can help you with or would love to chat about:

• Topics: Analytics in Production, Machine Learning, Deep Learning, Forecasting
• Languages: R, Python, SQL
• Fun: Hikes, Coffee

Contact Me

If you want to chat about work, collaboration, or just say hi, feel free to email me.

Visitor Count

map-of-tiny-perfect-things's Issues

Chatbot ideas

• Experiment with putting the information as a chat context instead of embeddings search
• Use GPT-4 instead of GPT-3.5

Nice-to-have Features

Brain dump of features we are considering

Front-end

• Instead of popup, show the details on the left 30% - bottom or top

Chatbot

• Idea: integrate chatbot as a popup inside the map
• Chatbot's first message should be the information (specifically the cities) included in the data

Crowdsourcing:

• Add checks for people when adding places (Google Ratings, meaningful descriptions)
• Make the Google Forms checking be automated based on our criteria

Outreach:

• make a substack
• share project in related forums (Pudding, Data Viz Slack)

Social Media:

• Location-based social network (meetups) (user log in system)
• Save Button

Descriptions:

• ChatGPT description generated from Google reviews

@harshvardhaniimi

Creator's Rec: how does it work?

When we created the maps' initial data, we "favourited" our places by calling them Creator's rec. When we make it crowdsourced using Google Forms to collect data, we cannot have them as creator's rec. Thus, data from Google Forms will never have creator's rec.

Is this how we want it to be? Do we want to create a "backdoor" where places can be made Creator's Rec afterwards?

Corollary to this is that new places that we (Dea and I) add will also not have Creator's rec.