Customer segmentation scbcn17
2 likes889 views
This document provides an overview of using customer segmentation and clustering algorithms for machine learning. It introduces the speaker, Julio Martinez, and his goal of learning and applying machine learning, using customer segmentation at his company Ulabox as an "excuse" to do so. It then provides a high-level overview of clustering algorithms like k-means and DBSCAN, the importance of data preparation, and explores clustering customers' ordering data to group them by behaviors.
![● Data preparation
○ Keep same order of magnitude, usually [0,1]
○ Remove noise
○ Other processes
■ Binarize data, categorical features
● weekday, ex. 4 -> 0, 0, 0, 1, 0, 0, 0
■ Process missing data
Before algorithms: data!](https://image.slidesharecdn.com/customersegmentationscbcn17-171006211733/85/Customer-segmentation-scbcn17-17-320.jpg)
Customer segmentation scbcn17
- 1. Customer segmentation an excuse to use Machine Learning ;-)
- 3. ● Julio Martinez ● Web developer since 2001 ● 2 years working at Ulabox ● Machine Learning hobbyist ● Find me: @liopic Who am I?
- 4. 1. docker pull jupyter/scipy-notebook 2. git clone git@github.com:ulabox/datasets 3. git clone git@github.com:liopic/scbcn17-customer-segmentation 4. cp datasets/data/*.csv scbcn17-customer-segmentation/ Preparing the workshop
- 5. My 2017 objective: M.L. ● Motivation ○ It’s the new hot thing ○ AlphaGo beat Lee Sedol, March 2016 ● Some background, but need to learn more
- 6. 1. Choose the way ○ Coursera’s vs. books vs. workshops vs. posts 2. Find an excuse to apply it ○ @work is better than @home Learning about Machine Learning
- 7. Customer clusters @work, aka “the excuse” ● There is a non-programmer Business Analysis Department ● Groups of customers based on periodicity + amount spent ○ Example: people that buy once per month, 100€ ticket ○ Useful for business reports ○ Not so useful for UX, CRM ● Groups by behavior? Clustering orders! Boring!
- 8. 1. With past data -> make a ML model ○ clean data ○ choose a ML algorithm/s ○ tune the algorithm, with testing 2. With new data -> use model to predict (or give new info) ○ deploy pipeline ○ update model 101 Machine Learning: the method
- 9. ● Supervised ○ data + labels(result) ● Unsupervised ○ just data ● Reinforcement ○ function to optimize 101 Machine Learning: type of problems
- 10. Supervised learning TRAINING SET cat cat person TEST SET ???
- 11. Unsupervised learning TRAINING SET TEST SET There is NO test
- 12. ● Try to extract features (information, shapes): similar and different ● Uses: ○ Clustering ○ Anomaly detection (it doesn’t look “normal”) ○ Dimensional reduction ○ Transfer features, projections ... Unsupervised learning
- 13. ● Use: ○ grouping ○ quantization ● Algorithms: ○ k-means ○ DBSCAN Clustering
- 14. ● need: how many clusters k-means
- 15. ● need: how many samples at minimum, tune other params DBSCAN: Density-based spatial clustering of applications with noise
- 16. So, ready to hack? But wait a moment!
- 17. ● Data preparation ○ Keep same order of magnitude, usually [0,1] ○ Remove noise ○ Other processes ■ Binarize data, categorical features ● weekday, ex. 4 -> 0, 0, 0, 1, 0, 0, 0 ■ Process missing data Before algorithms: data!
- 18. ● Explore the data ○ Images are richer than numbers ■ “We get more orders at 22h” vs. ● Ask domain experts ○ Understand normal & border cases ■ The step at 14h is the web cutoff time Before algorithms: data!
- 19. ● Explore and optimize the data ○ Features that count, feature engineering ○ Avoid the “curse of dimensionality” ● Start small, understandable, useful ● Find excuses to try it, and sell it! Lessons learned
- 20. Now, let’s hack!
- 21. 1. docker pull jupyter/scipy-notebook 2. git clone git@github.com:ulabox/datasets 3. git clone git@github.com:liopic/scbcn17-customer-segmentation 4. cp datasets/data/*.csv scbcn17-customer-segmentation/ 5. cd scbcn17-customer-segmentation 6. ./jupyter.sh 7. Open the link in your browser and open the Workshop.ipynb file Let’s hack
- 22. Thank you!