online-shoppers-intention

Using SAS Enterprise Miner to classify site visitors to be positive or negative buyers.

Project Intro

Sampling, Exploring, Modifying, Modeling, and Assessing (SEMMA) with SAS Enterprise Miner. Experiment with various methods to explore and develop models for the classification.

Methods Used

• EDA
• Resampling
• Clustering
• Segment Profiling
• Decision Tree, Neural Network
• Evaluation

Technologies

• SAS Enterprise Miner

Project Description

1. Sampling
   Data source from [UCI Machine Learning Repository]. Data class is imbalanced with 86% of negative cases.

2. Explore
   • Clustering method is used in this project to segment the data.
   • The Segment Profile tool is used to examine clustered data and identify factors that differentiate data segments from the population.
   • Chi-square test shows variables PageValues and ExitRates ranked the highest.
3. Modify
   • Replacement of missing values.
   • Resampling of data.
4. Model
   • Decision Tree
     
5. Assessment
   • Decision Tree ranks PageValues as the most important variable
     
   • Model Comparison
     

Getting Started

1. Download all the files and start a new SAS Enterprise Miner project.
2. Click on File and select Import diagram from XML.
3. Import Online Shoppers Intention.xml.
   
4. Select the File Import node. To specify the path click on Import File and browse to the location of online_shoppers_intention.csv.