NICF- Text Analytics
Do you know how to analyse customer sentiments about your company, products and services? Or how to keep track of your company’s service & quality delivery so that you are able to act quickly to insights that drive your business?
About 80% of enterprise-relevant data is in unstructured or semi-structured format. These include emails, documents, surveys, feedback forms, warranty claims, contact-centre notes and transcripts, web pages, news, data from social media, audios, videos and many more. A predominant amount of such data is available as text.
In order to gain competitive edge in the market, businesses and organisations are finding a growing need to expand their analysis scope to cover text data, especially in regards to customer feedbacks and social media data. This is so that critical insights can be uncovered to support business decision making and process improvement.
This course aims to equip you with the knowledge and skills to effectively analyse large amounts of textual data such as customer feedbacks and social media conversations to discover themes, patterns and trends to aid in improving business process and decision making. In scenario-based case studies, you will be introduced to common text analytics tasks such as data pre-processing and preparation, linguistic/knowledge resources management, concept extraction, text categorisation, clustering, association and trend analysis. You will practise performing these tasks following a well-defined process in hands-on sessions.
This 3-day data mining course focuses on introducing the essential analytical skills in modelling unstructured textual data such as customer feedbacks, reviews or comments to business and IT professionals.
This course is part of the Analytics & Intelligent Systems Series offered by NUS-ISS.
At the end of the course, participants will be able to:
- Identify main themes or topics in the collection of documents or textual data (e.g. the prominent issues customers are complaining about).
- Discover relationships and patterns among topics (e.g. which issues tend to co-occur in complaints).
- Categorise documents based on discovered topics and user-definable criteria, such as grouping complaints about similar issues for further investigation.
- Perform sentiment analysis on customers’ comments, reviews, or other forms of opinions to gain a good sense about how customers feel about their company, products and services.
- Extract useful information from text as structured data to enable integration into the traditional data mining process.
- Incorporate business understanding and domain knowledge into the analysis through lexical and knowledge resources.
- Perform the above tasks using the open-source language
What Will Be Covered
- Identify text analytics solution and platform requirements with IT team
- Develop term-document frequency matrix to enable lookup of text and documents within the corpus
- Define the metadata and corpus for the data to be imported into the text analytics repository
- Develop a standardised set of text analytics artifacts with the relevant stakeholders
- Modify the text analytics solution to ensure that it produces the expected results
- Define the process to perform text analytics based on the business requirements and text analytics artifacts
Lectures and workshops
Who should attend
This course is designed for both Business analytics and non-business analytics professionals. These include:
- Business and IT professionals seeking analytical skills to handle large amounts of textual data (e.g. customer feedbacks, product reviews on social media, etc.) for insights to improve business process and decision making
- Individuals who have no knowledge or experience in text analytics and would like to gain some knowledge in this area so that they may explore work opportunities in business analytics
- Data analysts, business users and IT professionals who want to move from the structured data to large amounts of unstructured, text data
The course will be conducted using IBM SPSS Modeler. A basic understanding of data mining will be desirable.