10.20.2020 | Ross Metusalem
Watch Recording | PPT
Key Takeaways
- Free or unstructured text (e.g., customer comments) can be analyzed statistically just like other types of data
- Every industry has use for text analytics, even manufacturers
- Not everybody realizes what’s possible with text analytics, so we might need to help build the vision
- Text Explorer makes text analytics accessible to those without specialized backgrounds
- Majority of Text Explorer’s value is in JMP Pro
What is Text Analysis?
Finding informative patterns in free or unstructured texts like customer comments, clinical narratives, maintenance logs, tweets, etc.
Value Statement
Text analytics uses statistics and data mining to derive information from text sources so organizations can make business decisions more effectively.
Typical Audience
- Service industries: Analysts use customer comments, interactions, reviews to improve services and predict buying behavior
- Manufacturing: Quality teams use logs, warranty claims, customer comments to improve products and understand real-world use
- Life Sciences, Pharma: Researchers use logs, narratives, notes to assess efficacy, compliance
- Government: Analysts use public comments to assess opinions, reports or narratives to characterize or predict real-world situations or outcomes
Problems/Pain Points Addressed by Text Analysis
- Text might seem impenetrable to statistical analysis, but text analytics makes it possible to analyze text like other data
- Text analytics allows processing of text at a scale and consistency not possible via human reading
- Text analytics can find information in text that might never be found otherwise
Terms/Phrases to Listen For
Types of Data:
- “free”, “unstructured”, “natural”
- Written reports, logs, notes
- Surveys
- Free response
- Narratives
- Claims
- Reviews
- Comments, complaints
- Posts
- Calls, chats, transcripts
- Verbatim
Types of Needs:
- Keywords, phrases
- Text (or word) similarities, differences
- Themes, topics, subjects, concepts
- Group / flag / label texts
- Opinions, feelings, sentiment
Technical Terms:
- Text mining
- Natural Language Processing (NLP)
- Word cloud
- Document-Term Matrix
- Document classification / clustering
- Latent Semantic Analysis
- Topic Analysis
- Sentiment Analysis
Discovery Questions to Ask
- Do you have any written content or text that you want to analyze statistically?
- Who (or what) generates the text? What does it contain? Can you provide an example?
- What do you wish your text could tell you?
- Are you analyzing your text data now? How?
JMP vs JMP Pro
- JMP finds keywords and phrases, produces word clouds
- JMP Pro has text mining techniques for finding more complex patterns
- JMP Pro 16 will have Sentiment Analysis for assessing positive-negative opinions in written comments
Related Resources
