Recommendation Engines In Python

-

Talk for PyData,Delhi 2016

K A J A L

 

Introduction to Recommendation Engines

• Definition
• Applications
• Various frameworks other than python
• Why Python is best suited for it.

1. Better for customization

2. Parallel Computing

Types of Recommendations

• User-User Based
• Item-Item Based 
• User-Item Based

Step Wise Guidance

• Data Acquisition (from databases usually)
• Data Cleaning (extracting out useful data)
• Shaping the data for various algorithms

Python Libraries Used

• pysuggest
• NumPy
• SciPy
• scikit-learn(sklearn)
• scipy.spatial
• Crab (Python framework for building recommender engines integrated with the world of scientific Python packages (numpy, scipy, matplotlib).

Problems Faced

• 80% of the time will be spent on data gathering and cleaning it for training purposes.
• Differentiating which algorithm will fit to your data set in the most efficient way