Design and Implementation of Item Based Collaborative Filtering -A Case Study

Data centric intelligent applications are acquiring predominant importance due to their ability to churn data and provide meaningful insights for decision making. Machine learning techniques, especially recommendation systems built using collaborative filtering are quite commonly being used in online retail industry applications and e-commerce industry. These algorithms can also be leveraged even in the other domains. In the present work, item based collaborative filtering technique is used and implementation aspects are discussed at abstract level, specific to the use case of personalized capability development requirements across the projects. Each technical skill is considered as item. CRISP-DSM framework is followed and Loglikelyhood similarity measure is used leveraging Apache Mahout framework. It provides insights into identifying and recommending skills for business based on the past trainings requirements.