class Prabhat:
def __init__(self):
self.role = "Third-year B.Tech CSE Student"
self.university = "Galgotias University"
self.location = "New Delhi / Bihar, India"
self.focus = "Ethical Hacking & Cybersecurity"
self.currently = ["Python", "Java", "Data Structures & Algorithms", "Bash"]
self.building_in = "Open Source + Data Science"
def say_hi(self):
print("Thanks for stopping by โ let's build something!")
me = Prabhat()
me.say_hi()
|
|
๐ต๏ธ Click to reveal a fun fact
I'm most productive at 2 AM debugging something that worked perfectly fine five minutes ago. ๐โ
Fully designed and built end-to-end โ not a fork, not an issue fix.
Analyzes real e-commerce transaction data (the UCI Online Retail dataset) to understand customer buying behavior end-to-end โ cleaning, EDA, RFM (Recency, Frequency, Monetary) analysis, K-Means clustering, and a live Streamlit dashboard that turns raw transactions into segments a marketing team could actually act on.
๐ Click to expand: how it works
| Stage | What Happens |
|---|---|
| ๐งน Data Cleaning | Handles missing values, duplicates, and type conversions on raw retail transactions |
| ๐ EDA | Visualizes purchase patterns, order frequency, and revenue distribution |
| ๐งฎ RFM Engineering | Computes Recency, Frequency, and Monetary scores per customer |
| ๐ฏ Segmentation | K-Means clustering (Elbow Method + Silhouette Score) groups customers into segments like Champions, At Risk, and Hibernating |
| ๐ Dashboard | A Streamlit app makes the segments explorable interactively |
Debugging note: this project went through multiple audit cycles across the pipeline, the app, and the notebooks โ including tracking down a sneaky .gitignore issue that was silently blocking deployment to Streamlit Community Cloud.
|
๐๏ธ civicFix Snap a photo of a pothole, broken streetlight, or overflowing bin โ tag it, put it on a map, and let the community upvote it so authorities stay accountable. |
๐ฝ๏ธ SmartDineQR Scan, order, and pay at a restaurant table with no waiters and no printed menus. |
