BUSINESS INTELLIGENCE APPROACH TO MARKET RESEARCH ON FOOD COMMODITY BY USING BIG DATA ANALYTICS, CASE STUDY: FORUM JUAL BELI KASKUS

• Main Author: Wibowo, Andi
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63474
• Institution: Institut Teknologi Bandung

The emerging Covid-19 pandemic and the increasing use of internet access trend in Indonesia have successfully changed most consumer behavior, shifting into the online market (e-Commerce) instead of the conventional physical market. It also presents vast new business opportunities, particularly in the food and beverage industries. However, the current traditional market research known for the difficulties of finding and collecting meaningful data in online market. Additionally, inconsistent manual data gathering is another factor contributing to the inefficiency of traditional market research. The rapid digitalization of social life and commerce has resulted in consumer and business digital footprints which a valuable source to begin market research. The digital footprint from public access dynamic website like e-commerce platform, could be transform into a knowledge by implemented the Big Data Analytics. Utilize Big Data analytics can sift through massive databases to obtain actionable insights translated into organizational strategy and decisions. This research aimed to present the Big Data Analytics power to find insight from the current condition of e-commerce by conducting market research on Forum Jual Beli Kaskus, which was chosen due to less restriction while implementing the self-programming web scrape engine. The research conceptual framework including four analytics step including : The Business Issue Analysis, Data Collection using Java Programming Language for self-programming web scrape engine, implementation of text mining and machine learning for predictive model (using Rapidminer software), then visualization through business intelligence tool (using Tableau for Desktop software). Authors find that the web scraping engine successfully collected the whole population of products listed in the food and beverages category as well as text preprocessing resulted in several keywords which represent the product trend. The prediction model (machine learning) achieved 99.85% accuracy and a minimum 80% precision class while the test dataset was introduced to confirm and test the model. From the analysis of the application and model test results on 100 food and beverage product data obtained from various kinds of e-commerce, it is known that the prediction model built to determine potential law violations is able to accurately classify 78% of product and another 22% are missed. The analysis result from web scraping, text mining, and machine learning is visually translated into the dashboard as a Business Intelligence Tool. The machine learning model above could bring further research with diverse data and complex words to increase accuracy and precision. Also, the model could be developed by implementing the other parameters like product price, seller location, seller rank, and many attributes previously collected by the Web Scrape engine. By utilizing The Big Data Analytics model and Business Intelligence Tools, the government or authority could catch enormous insight based on the real-time market research process to formulate a better policy approach.