A Random Forest Approach for Predicting Online Buying Behavior of Indian Customers

摘要

Online retailing in India has shown remarkable growth in the recent years. Despite having a low internet penetration rate of 34.5%, India has the second largest number of internet users in the world after China. Given the growing importance of the online retail industry in India and its diverse set of sensitivities and region wise socio-psychological barriers, it is imperative for retailers to understand customer shopping preferences. In this paper, we attempt to understand various factors influencing the online buying behavior of Indian customers in different product categories, across geographic locations in India. Also, we developed and validated the Random Forest prediction model for each identified product category, to understand if the Indian online shopping market is ready for these product categories or the traditional channel is preferred over by customer. A questionnaire based survey is used to collected data from 124 Indian respondents from 18 states of India. The survey captured from both offline and online shopping environment to aggregate understanding of customers’ shopping preferences. The high Sensitivity (above 85%) of the Random Forest model for Books and Electronics categories suggests inclination of purchase intension of customer towards online shopping. Retailers can use this model to predict the buying behavior of customers based on the location. However, for product categories like Movies, Sports equipment and Handbags lang="EN-US" style="font-family:Verdana;">, lang="EN-US" style="font-family:Verdana;"> the high value of Specificity signifies the model prediction towards offline purchase intensions. So for these product categories retailers may like to focus more on customer services at retail stores.