All Collections
Getting started
How do I export the results of my analysis?
How do I export the results of my analysis?

Learn how to export the results of a comparison into a spreadsheet file and interpret the various metrics.

Trish Pencarska avatar
Written by Trish Pencarska
Updated over a week ago

You can easily export the raw data behind your analysis from the Relative Insight dashboard. 

Exporting the results of a comparison

  1. From the comparison explorer, use the filters to define what you want to look at

  2. At the top-right of the table, click the export icon

  3. Choose the desired format: CSV or Power BI to download the file.

Understanding the report

Exported files will include all raw data associated with the comparison whereas the results presented to users in the platform only include findings that meet the applicable significance threshold.

Tag & description – the word, phrase, topic tag code, grammar tag code, or emotional category 

Frequency – the number of occurrences of the item within your primary data set (language set 1 freq) and comparison data set (language set 2 freq)

Relative frequency – the frequency of the item in your primary (language set 1 relative freq) or comparison (language set 2 relative freq) data set, divided by the total number of words in the data set

Relative difference – The relative frequency of your primary data set divided by the relative frequency in the comparison data set, a measure of how much more prevalent the feature is in one data set compared to another

LL value – the log-likelihood significance score. Positive values indicate that the item occurs significantly more in the primary data set, and negative values indicate it occurs more in the comparison data set. The greater the value (positive or negative) the stronger the confidence.

Frequency vs message frequency

The values displayed in the platform are based on the number of instances of a particular linguistic feature in comparison to the overall size of the data set (word count). If a particular feature occurs multiple times within a single block of text (e.g. a single survey open-end, customer review or tweet), each instance will constitute a unique mention.

The message frequency metrics included in the analysis export look at how many individual blocks of text (messages) contain a particular linguistic feature. If a linguistic feature appears multiple times within a single message, it will be treated the same as if it only appears once.

The message frequency approach helps to minimize the impact one piece of text can have on the overall analysis (e.g. a single berating customer review filled with expletives).

Did this answer your question?