Use data Mining Techniques to Predict Users’ Engagement on the Social Network Posts in The Period Before, During and After Ramadan
The British University in Dubai (BUiD)
Social media has become ubiquitous and main requirement for individuals lives, its used for socialization sharing information and recently for marketing and advertising. The content of the posts has a significant impact on attracting users and total user’s engagement. Companies have adopted social media as an essential tool for realizing potential customer needs and to ensure optimal customer satisfaction. Analyzing the content, type and the best time of posting can effectively impact and benefit a business. Data mining (also referred to as Data Knowledge Discovery) is the practice of examining a dataset to establish the hidden patterns and knowledge in order to represent the results in an understandable format. In this thesis, the effectiveness of using a data mining classifier helps to predict users’ engagement with a social media post before publishing and anticipating the best type of media for the post. Different classification algorithms were applied to the dataset using the Rapidminer tool. The results of the classification considered as a baseline for this research. In contrast to traditional approaches on this topic, this dissertation seeks to analyze the depth of users’ engagement with social media posts and discuss the basis for a predictive model that to predict the total engagement of a post before publishing. At the beginning, a data set was collected from Crowdbabble online tool and the collected dataset presented different periods and different social media networks to examine the user behavior at different times. Before Ramadan presented a typical month, During Ramadan related to the special month for Muslims and after Ramadan referred to the beginning of summer vacation. Facebook, Twitter and Instagram were the main platforms examined as these are popular in the Arab world. This study focused on dataset from the Arab world to investigate Arab users’ behaviors and interests. The collected dataset was analyzed from various perspectives to study the post characteristics and the effect on users’ engagement. Each platform (Facebook, Instagram, Twitter) has a primary role in users’ engagement, Facebook is the top used application tool compared to Instagram and Twitter. The results showed that Post type (Status, Photo, Link, Video) has an important role in attracting users and the analysis confirms that video and photo posts create the maximum levels of engagement. Post fields also had impact on engaging users, whereby the categories of Beauty, Fashion and Celebrity were the most attractive page types.
data mining, users’ engagement, social network posts, Ramadan, social media