Customer Segmentation in Python
summary
Walks through concrete Python code (pandas, scikit-learn, scipy) on open booking data: loads CSV, cleans anomalies, computes booking rates per channel (Channel 0 ≈ 12,000 attempts, 7.2% vs overall 7.9%), implements a custom st_comparison function to get z-scores and p-values and flags significance at p<0.1. Runs k‑means (k=3 clusters), uses PCA to plot, merges cluster means to label one cluster as long‑stays and another as family groups, trains decision trees with an 80/20 split and finds a subgroup with ~30% booking vs 8% baseline.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 2:10Loading the dataset
- 6:35Validating business logic
- 9:55Underperforming segments
- 12:45Statistical significance testing
- 16:40Clustering cities for marketing
- 25:26Booking rates with decision trees
- 31:36Summary and next steps
- 32:17Q&A
related talks
15mMachine Learning with Python Case Study
24mSegmenting Customers with Machine Learning
30m23, Python and Me: Using Machine Learning in Python to Analyze Consumer Genomics Data
34mHow to understand user behaviour using Multiple Linear Regression?
42mFlying Python at 40,000 feet
01h 01mWorkshop: Intro to topic modeling using Python
24mPython, Pandas and other animals to survive in the data jungle
34mDimensionally reducing data squeezing out the good stuff