All Collections
Frequently Asked Questions
How do I discover Emergent Concepts?
How do I discover Emergent Concepts?

How to discovery trending concepts within your data.

Kris Rogers avatar
Written by Kris Rogers
Updated over a week ago

Our Emergent Concepts chart lets you easily discover the concepts that have changed the most over a specific time period, revealing important trends in your data. The chart itself appears on multiple pages in Kapiche, so we'll first give a general rundown of how to use the chart before discussing how it is used on each specific page.

Contents

Emergent Concepts Overview

By default, the Emergent Concepts chart will show concepts that have changed the most during the past 30 days. You can change the time period to a preselected option or enter a custom range using the dropdown.

You can also configure the minimum threshold for concept frequency. By default, we have set this to 30. This means that a concept has to appear a minimum of 30 times across our datasets' full time period. Adjusting the minimum threshold allows you to filter out low frequency concepts, to quickly surface concepts that are more meaningful to you.

If you have no data covering the selected period you will see a message indicating that the chart could not be created. We will be using a custom range for this example. Below is a snippet of the top concepts by change in % frequency.

How is change calculated?

Hovering over a concept's row will reveal details about the change it underwent, as shown below.

Given our selected date range -- 1 Jul 2021 to 30 Jun 2022 -- we calculate frequency change as follows:

The absolute (#) frequency change, which in this case is 268, is simply the number of times that concept appeared in the date range.


​The percentage (%) frequency change, which in this case is 2.52%, is determined by subtracting the percentage frequency of the concept up to (but not including) the beginning of the selected date period from the percentage frequency of the concept up to and including the end of the selected date period.

For example, in this case, the frequency of refund was 11.62% up to but not including the start of our date period (1 Jul 2021) and 14.14% up to and including the end of our date period (30 Jun 2022). Therefore, the percentage frequency increased by 2.52% (14.14% minus 11.62%) during the selected date period.

This pair of values reveals how much the concept increased in frequency both in absolute terms and relative to the total volume of data during the specified time period.

What is relative change?

You'll also notice in the tooltips a statistic called 'relative difference'.

Relative change helps to gauge the magnitude of change proportionate to the average of the two frequencies and is calculated as follows:

This metric can be useful in highlighting concepts that have undergone significant change, regardless of how large their overall frequency. Below we illustrate by showing that dropouts has a higher degree of relative change than days despite a significant lower change in frequency. In this sense you can think of relative change as measuring the acceleration of change for a concept.

Chart sorting options

By default the chart is sorted by Frequency %, in decreasing order, emphasizing concepts increasing in frequency. If you click on the Freq (%) header, it will reverse the sort order and show concepts decreasing in frequency. You can also click on the Freq (#), which will show the concepts with the greatest increase or decrease in raw frequency.

Unmapped Verbatims

When the Emergent Concepts chart appears on the Unmapped Verbatims page, it shows concepts with the biggest change based only on the subset of data that is not already captured in a theme. Because of this, you are more likely to detect the beginnings of new, unique themes rather than changes within existing themes that you are tracking. The theme drilldown of the Dashboard is the best place to achieve the latter.

Dashboard Overview

When the Emergent Concepts chart appears on the overview page of the Dashboard, it shows concepts with the biggest change based on all data available to the dashboard. When filters are not applied, you will see concepts with the biggest changes across the whole dataset showing you emerging trends regardless of whether they overlap with existing themes or not.

By applying dashboard filters you'll change the data used by the Emergent Concepts chart, allowing you to pinpoint key changes within a specific customer segment.

Dashboard Theme

You'll also see the Emergent Concepts chart when drilling down into a theme on the Dashboard. Like with the overview page of the Dashboard, filters will allow you to examine a particular segment of the data. In addition, the Emergent Concepts chart will be restricted only to the data that matches the theme being examined. This allows you to monitor new, emerging facets and changing discourse for a particular theme.

Questions? πŸ€”

If you have any questions about Emergent Concepts (or anything else!) you can get in touch with us any time by hitting the blue chat button to your right πŸ‘‰

Did this answer your question?