Banking & Financial Services
Banks (or customer service department of large companies), receive thousands of complaints on a daily basis and thus it becomes increasingly difficult to go through all these complaints and connect with the customer. Additionally, the complaints are often wrongly classified and it becomes time consuming to go through each complaint and redirect it to the relevant departments and these slow processing times also leads to slower response times to the customers. Also, the complaints which have higher priority and should be addressed immediately get sidelined due to lower priority complaints in the queue. Thus, it is critical for banks and large companies to have a customer complaint prioritization application to increase their response times and address higher priority customer complaints sooner.
We developed a POC using Natural Language Processing (NLP) to classify and prioritize complaints. Various NLP techniques like stop word removal, lemmatization, stemming, vectorizing using TF-ID, etc. were applied to pre-process the complaint data. Once vectorized, various machine learning classification algorithms like Logistic Regression, Support Vector Machine, etc. were used to classify the data into the correct categories automatically. For prioritization, Sentiment Analysis techniques were applied. Priority was assigned based on the sentiment and polarity of the text.