All of our widgets follow the same pattern. You use the Options tab to design your widget, for example by picking the variables you want to chart, and then the Filters tab to filter for a specific segmentation, if needed.

The Time series widget follows the same principle but is a touch more complicated, so a more in-depth explanation is in order.

The first thing you need to choose when setting up a time series is how you want it grouped.

Would you like to chart a specific metadata variable over time? If so, choose “A field”. This allows you to group by a chosen metadata field in a single dataset. You can choose the field from the “field” dropdown that appears once this option is chosen. You are then asked to choose the date field you want to chart by. Some datasets have more than one date field.

Next, choose the values you wish to calculate by, these vary according to the dataset chosen. You can add up to two calculated values, by selecting the +Add button.

Below we see the number of records, which in this example are public reviews of the UberEats app, from 4 cities during the month of March 2021.

You can name your widget on the top left, and on the top right side of the dialog you can choose the sorting order, the date range, and the level of granularity.

Choosing “Overall” allows you to chart a time series based on Keatext-generated metadata across multiple datasets. This provides one total, across datasets, over time, of the chosen Keatext-generated metadata variable. Keatext metadata is the same across datasets. It includes comment count, sentiment score, praise count, problem count, question count, records count, and suggestions count. Phew! That's a lot of words to say, for example, “total X count/score across all datasets over time.” ;)

Below I have two datasets and have charted the record count of both datasets combined.

“Sources” is similar to “overall” in that it charts a time series based on Keatext-generated metadata, however with this option it is broken down by the different sources. So for example if you have three sources you will have three lines on the chart, whereas by choosing “overall” those three sources will be combined into one line. Make sense? Good!

Below we can see the record count over time of two datasets taken from public online reviews plotted separately, one from the Better Business Bureau and the other from Indeed.

“Tags” only works if you have applied tags to your dataset. To find out how to do this read here. This option allows you to chart your tags over time using Keatext-generated metadata.

We tagged public UberEats reviews by customer journey step. Here we can see a time series of four of these steps. To display more than four choose a number from the “series per page” dropdown at the bottom of the dialog. You can also navigate between the pages with the pagination controls in the bottom right.

Keatext analyzes your data by finding the topic in a comment and its corresponding opinion. These topics and opinions are grouped with others of similar meaning. For example the opinion group “dirty” may contain words of similar meaning like "filthy, grubby, gross, not been cleaned”. Chances are if you have gotten this far and are designing dashboards, you already know this, but if you need a refresher on how Keatext analyzes your data you can find it here.

The good news is you can chart a time series based on chosen Topics or Opinions.

Below, using our public UberEats reviews dataset, I have charted the topic of “fee” overtime by record count and sentiment score. To choose the topic you want to chart, move to the filters tab and search for the desired topic in the search filter.

Ditto for opinions!