01 Establishment of appropriate inventory prediction model
Currently, small business owners maintain inventory through simple statistics and sense of sales volume so far. In order to pursue stable business maintenance, we applied the AI model to the inventory management.
While applying various models such as arima, multiple linear regression, random forest regression, lstm, and q-learning, we studied the characteristics of each model and the understanding of the characteristics of the data.
02 Interconnection data analysis and visualization
Using R to analyze CAIDA data and identify domestic and foreign interconnection data trends by quarter from 2004 to 2021
Comparing and visualizing the Korean-Japanese interconnection network structure, and gaining experience in identifying the characteristics of each country's network.
Based on the contents, a report is prepared and published as an ETRI internal report.
03 Classification of production activities of business entities NLP
Applying AI technology based on the relationship between natural language data that describes the production activities of a business owned by the National Statistical Office and the industrial classification of the business
It went through preprocessing of typos and keyword replacement of natural language data, and achieved a final accuracy of 89% by industry classification using the koBERT model.