Sunday 10:45 AM–11:30 AM in Room #1025 (1st Floor)

Triaging Feedback Form Data 

Stephanie Kim

Audience level:
Intermediate

Description

This talk will cover how to use predictive modeling on unstructured text data including feedback form, social media or chat message data to triage issues in order to prevent future problems with a service, platform or user interface using NLP techniques in Python and R.

Abstract

Companies gain useful insights about their users from feedback form and other unstructured text data including live chat messages. Even though they are read and responded to, often such data is ignored when thinking about larger scale trend analysis and this can result in missed insight about how users react to a product or service. Sometimes analysis is being done by looking at changes in user sentiment or other heuristics, however it could be taken a step further by applying predictive modeling in attempt to recognize areas that need more attention and support. While you can use predictive modeling on network and log data, that is looking at how the hardware is handling your users requests, not how it's being perceived by users. By predicting areas where users are having difficulty whether it's with the UI or with the platform's response time you can triage these areas of concern to prevent future cases of negative perception. This talk will cover how to utilize common NLP tools used to gather and process the features in Python then will use R to perform trend analysis and predictive modeling then use the results to triage what areas should be focused on in the future.