/lookalike-modelling

Finding customer lookalikes using Machine Learning in PySpark

Primary LanguagePython

Lookalike Modelling

Versions:

Spark: 2.1.1, Python: 3.5
  • Objective of the project is to find customer lookalikes (similar customers) in a large population using algorithms like Random Forest, KNN and KD-Trees
  • Cannot share data as it is sensitive to clients but need to follow the following format:

Train Data:

col 1 -> ID

col 2...n-1 -> features

col n -> Label (Binary)

Test Data:

col 1 -> ID

col 2 -> features
  • The train code will treat 1 as minority class and generate the 30% customer IDs that are lookalikes after running predict.py
  • Please manually fill the file locations in the beginning the code for input files, intermediate files etc.