This article explains how aggregation options work.
The time series chart, categorical chart, radar chart, matrix, aggregation table, tile and pie/donut widgets allow you to aggregate the data that is displayed within these widgets. You can use the aggregation options to decide which record or which value from your data source should represent the field you’ve chosen to visualise.
Aggregations work slightly differently depending on the type of widget you’re using them in.
In a pie/donut chart or a categorical chart, the aggregation is performed for all values that fall within the category or categories that you’re charting. So if you were visualising injury data for a team, you might use the injury type (chronic, acute) as the category, then chart the sum (aggregation) of the injury duration field. This would get you a chart showing how many days the team has lost to chronic injuries versus acute injuries.
For a time series chart, the aggregation is performed for all values that fall within each of the dates covered by the settings used to define your data source. For example, you could use the mean aggregation to understand how a team’s average training load changes over the course of a month. Each point on your time series chart would then be the team’s mean training load for that date.
If you’re working with a tile widget, then you can aggregate the single series which is displayed as the tile value.
Within a radar chart widget, you can aggregate each series that is plotted on the radar chart.
When you’re using a matrix widget, you can aggregate the values that appear as cells in the matrix. You can also aggregate any summary rows or columns that you add to the matrix.
If you’re using the aggregation table widget, you can aggregate the category values that appear in the first column of the table. You can also aggregate the values that appear in each subsequent column of the table.
Aggregations for number type data
- Sum: returns the sum of all values contained in the dataset.
- Mean: returns the mean value for the dataset.
- Median: this returns the middle value in the distribution of the dataset.
- Standard deviation: this returns the square root of the variance of the dataset. This is useful for understanding how spread out the data is.
- Mean +/- n standard deviations: this returns the mean from your data source plus or minus the specified number of standard deviations.
- Nth minimum: this returns, for example, the fifth smallest value from the data. You can also use the option to specify unique values only should be included in the aggregation process.
- Nth maximum: this returns, for example, the 10th largest value from the data. You can also use the option to specify unique values only should be included in the aggregation process.
- Quantile: the quantile determines how many values in your dataset are above or below your input (a value between 0 and 1). For example, you can use 0.45 to return the value where 45% of the data is lower than that value and 55% is higher than that value.
- Variance: returns the variance of the data.
Aggregations for number and string type data
- Count: returns the total number of records contained in the dataset after any filters are applied. You can also use the option to specify unique values only should be included in the aggregation process.
Aggregations for number and date type data
- Minimum: returns the smallest value from the data source. If field you’re using is a number, you’ll get the smallest number. If the field is a date, then you’ll get the earliest date.
- Maximum: returns the largest value from the data source. If field you’re using is a number, you’ll get the largest number. If the field is a date, then you’ll get the final date.
Aggregations for all data types (date, image, number, string)
- Get from most recent record: returns the data from the record with the largest date and time value. When working with future data, the most recent record is always the final record.
- Get from earliest record: return the data from the first record from the dataset defined by the data source. When working with future data only, the earliest record is always the record closest to the current date.
- Get from nth earliest record: return the data from, for example, the second record of the dataset.
- Get from nth most recent record: return the data from, for example, the record with the third largest date and time value.