Abstract 

The International Society for Krishna Consciousness (ISKCON), also known as the Hare Krishna movement, was founded by His Divine Grace A. C. Bhaktivedanta Swami Prabhupada in New York City in1966.  ISKCON has 850 temples and centres worldwide.  ISKCON, Bangalore attracts hundreds of visitors every day and ISKCON’s IT department collects feedback from visitors about their experience at ISKCON.  In addition, they also collect comments written by the visitors in social media platforms such as Facebook and TripAdvisor.

Janarthanan Balasubramanian, Division Head, Information Technology and online Communications at ISKCON wanted to understand the visitor feedback so that appropriate measures can be taken to improve the visitor experience at ISKCON.  The primary problem at hand for Janarthanan was to reduce the existing manual effort for his team. Currently three resources are involved in collecting the reviews from social platforms and labeling each review into one of the four classes, viz. positive, negative, neutral and mixed. Two other resources convert the reviews from paper feedback forms / feedback registers placed at different points inside the temple, to an Excel file. The team begins its day by manually collecting, labelling and collating the reviews in an Excel file. At the end of the day, these labelled reviews would be stored in the database. At the end of the week, the total count of reviews for the four classes, viz. positive, negative, neutral and mixed was calculated to understand the overall sentiment. This was an extremely time-consuming manual process from data collection, that is, manually copy pasting the comments from social mediums to data collation and labelling.

 Janarthanan wanted his team to spend time and effort on analyzing the data and working on remedial actions. He wanted to understand the issues/topics that ISKCON should work on, rather than manually classify reviews and get the count of each review type. Janarthanan felt that getting only the split of positive/negative/neutral/mixed classes is not enough to draw inferences.

Learning Objective (Maximum of 500 Characters):  Briefly describes teaching goals of case.

The case can be used for teaching text analytics and sentiment analysis and various challenges in analyzing text data.  The learning objectives of this case are as follows:

• Understand data issues associated with text data.
• Learn to perform exploratory data analysis (EDA) using text data.
• Learn to extract features from text data for sentiment analysis

Understand how to develop an appropriate model to predict sentiment of a comment.