Back to: Index of cards

Operations in this section include transforming existing time series into derived series and visualizing time series data on charts.

- Add time series
- Object time series property
- Time series preview (Beta)
- Numeric formula
- Time series formula
- Derivative
- DSP filter
- Filter time series
- Integral
- Regression
- Sample
- Segment transformation
- Union
- Relative time
- Time range
- Time shift
- Cumulative aggregate
- Rolling aggregate
- Periodic aggregate
- Events plot
- Event indicator series
- Time series search
- Batch time series
- Grouped time series
- Linear aggregation
- Bollinger bands
- Time series bound

Opens a pop-up menu where you can search and add time series objects to the analysis.

Get a time series property or linked sensor associated with an object. To add a linked sensor, follow the steps below:
1. Select a root object type (for example: `Weather Station`

).
2. Select a specific object instance (for example: `ROMA`

).
3. Select a sensor from the dropdown showing available sensors. Once a sensor is selected it is rendered as a time series plot.

Takes as input a Foundry dataset of a specific schema and returns a preview of the time series on a time series chart.

- The dataset must contain a
`Timestamp`

column of LONG type, a`Value`

column of DOUBLE type, and a`Series ID`

column. - The
`Timestamp`

column will be divided into 100 buckets and the`Value`

column will be averaged for each bucket. This will cause displayed values to be approximate.

Time series preview cards are a Beta feature and may not be available on your Foundry instance.

Generates a time series plot by writing a formula in the formula box.

In the formula box, use any of the assigned plot variable names to type in a mathematical formula returning a numerical value.

- Basic arithmetic operations are supported (multiplication, division, addition, or subtraction, for example), along with more complex mathematical expressions.
- Quiver’s autocomplete will suggest built-in functions along with existing series based on parameter name or series properties.
- Quiver uses a custom mathematical expression language for formulas. More details about supported expressions are available in the formula documentation.

The Derivative plot shows the rate of change at each given point on the selected plot.

- The default is to calculate the rate of change per second. This can be changed to several other options. Note that this does not affect the shape of the curve, only the y-axis units.
- Derivatives are useful for identifying when the slope of a series is flat (that is, not changing.) To find periods where a series is not changing, you can do a Time Series Search for periods when the derivative of the series is equal to zero.

DSP (digital signal processing) filters are commonly used on incoming series to reduce their complexity. Quiver includes three separate filtering algorithms: Butterworth ↗, Chebyshev ↗, and Inverse Chebyshev ↗. Each algorithm has several parameters that can be altered.

This filter allows you to keep or remove points based on mathematical conditions. While the Time Range transformation lets you filter data based on the timestamp of the data (the x-axis, in many cases), the Filter transformation filters based on the value of the data (the y-axis, in many cases). As an example, to filter a series to only include values over 100, you can write `$A > 100`

.

- Note that we use the
*plot reference*to refer to a specific series (`$A`

). This allows you to use multiple series in your filter condition (`$A > 100 && $B < 50`

). - There is a parameter to control whether to keep or remove the data that matches your filter condition.
- There is a
**Filter Method**parameter with two options:**Gap**and**Continuous**.- If you choose
**Gap**, data not matching your condition will be filtered out and will be shown as gaps in your series. - With
**Continuous**, Quiver will interpolate between the filtered values to produce a continuous series. Interpolation settings are controlled by the**Interpolation Options**dropdowns, with default**Linear**.

- If you choose

The Integral transformation is the reverse of the Derivative. Rather than calculating rate of change, it calculates the area under the curve.

- As with derivatives, you can calculate the rate over several different time units.
- In addition to linear integration, you can also perform LHS and RHS integration using the
**Integration Method**option.

Regression is used to view the best-fit regression line over a selected time series.

- Linear, polynomial (of degree 0-13), or exponential regression fits can be chosen.
- The time range used to compute the best fit line is set by default to be the source plot zoom range, but can be modified to match a time range parameter instead.

Sample is used to resample an existing series at a constant frequency.

- This can be used in two primary scenarios:
- Data is coming in at a constant rate (such as daily), but some days no data was recorded. Rather than having gaps in the data, you can use Sample Series to resample at a daily rate to produce a complete series.
- Data is coming in at a constant rate (such as daily), but you would like a series that has data at a different rate (such as hourly or weekly).

- Sample Series calculates its new points by using interpolation between the existing data. There are a variety of interpolation options available to sample the new points. The default method is
**Linear**interpolation, but**Previous**,**Next**,**Nearest**, and**None**can also be used. These are controlled by the**Interpolation Options**dropdown.

Segment transforms operate on the output of a Filtered series.

- Filtered series will often be shown as discrete segments, with gaps in between each segment. The Segment transformation allows you to calculate statistics about each segment (such as average, max, or standard deviation).

The Union transformation operates on two or more series and allows you to compute a running sum or mean of all inputs.

The relative axis plot type can be used to plot series against a time-axis that is not absolute. Instead, you can plot relative to the source plot used.

- In the example below, the relative axis is aligned to the start time of the series. Therefore, the X axis displays days/years since the first time point, instead of the absolute time.
- In addition to aligning to the source plot, you can also align to arbitrary time ranges, time series searches, or arbitrary custom dates.

Time range can be used to filter your series to a defined time range. It takes a time series plot and a time range parameter as inputs.

Time shift allows you to shift the time of a series by some time unit.

- Series can be shifted forwards or backwards.
- This is useful for looking at leading or lagging indicators.

Cumulative aggregates allow you to display the cumulative value of a series, either over the entire length of the series or over a specific period of time.

- For example, if we had a series representing the dividend payout of Disney stock over time, we could use a Cumulative aggregate sum to see the running total of dividend payout as it grows with time.
- There is also a
**Cumulative over Display Range only**toggle. By default, Cumulative aggregates will calculate over the entire range of the series. If you are zoomed in and only want to compute aggregations over the time range displayed, switch this toggle on.

Rolling aggregates are typically used to “smooth” a series and show an averaged version of it. For each point in your series, a Rolling aggregate will calculate a new point based on your window function and aggregate method.

- As an example, if you choose a window size of one week and method
`average`

, each point will be calculated by finding the average value over the previous week. - If you choose a window size of three days and method
`sum`

, then each point will be the sum of the previous three days. - Windows are calculated by previous values up to and including that point.

Periodic aggregates are similar to Rolling aggregates except that they downsample the data.

- If you have daily data and perform a Rolling aggregate using a window of one week with an average function, your chart will return a series with one point per day, with each point representing the previous week’s average.
- However, if you do a Periodic aggregate with a window of one week, your new series will have one point per week rather than one point per day.

Take an object set of time series root objects and a time series formula, and plot the formula for the full batch of root objects.

Display an object set of time series with the options to color and separate based on any object properties and slice the time series by linked events, filtering only to relevant time slices.

Takes multiple time series, either from a transform table column, a grouped time series plot, or an object set with time series, and computes a linear aggregation of the time series set over time. You can add a linear aggregation from the **Next Actions** menu by selecting **Visualize > Linear aggregation**. Note that while a rolling or periodic aggregate transforms aggregate over a single time series, the linear aggregation aggregates across multiple time series.

A Bollinger band plot is a type of technical analysis tool used by traders and investors to analyze and identify potential trends, price volatility, and market sentiment in financial markets. The plot consists of two bands plotted around a simple moving average (SMA) on a configurable rolling time window. These bands are:

**Upper Bollinger band**: This band is plotted at a multiple of standard deviations (typically 2) above the moving average.**Lower Bollinger band**: This band is plotted at a multiple of standard deviations (typically 2) below the moving average.

As shown in the example below, these bands are usually plotted together with the moving average itself and they can be added separately using the rolling aggregate plot.

The time series bound plots the region bounded by two time series.