Potential customer mining application of smart home products based on LightGBM PU learning and Spark ML algorithm practice
This work addresses a domain-specific problem for telecom companies by applying incremental machine learning techniques to improve customer mining for smart home products.
The paper tackled the problem of identifying potential customers for smart home products using a large dataset from China Telecom Shanghai Company, achieving results through the application of LightGBM, PySpark, and Positive Unlabeled Learning algorithms to predict customer purchases and enable precision marketing.
This paper studies the case of big data-based intelligent product potential customer mining internal competition in China Telecom Shanghai Company. Huge amounts of data based on big data table, the use of machine Learning and data analysis technology, using the algorithm of LightGBM, PySpark machine Learning algorithms, Positive Unlabeled Learning algorithm, and predict whether customers buy whole house product, precision marketing into artificial intelligence for the customer, large data capacity, promote the development of intelligent products of the company.