Knowledge Base

The Difference Between Select Intervals by External Data and Select Intervals by Manual Selection with External Data

With selecting intervals by external data, the upper and lower limits of the interval are fixed and based on the external data only. In the picture below, the intervals are based on a heart rate between 80 and 100 bpm. The selected intervals are shown in the white areas.

With selecting intervals by manual selection, you can include time in the interval start and end criteria, for example 1 minute after the heart rate becomes higher than 100 bpm. Furthermore, there need not be an upper and lower limit; you can use higher than or lower than for both the start and the end of the interval. In the figure below, the interval starts when the heart rate becomes higher than 100 bpm and stops when the heart rate becomes higher than 80 bpm.

Note: Depending on the sample rate of the external data, in the event plot it can sometimes seem as if the interval does not include external data values that should have been included. This is because in the event plot samples are connected by a line as if the external data is continuous, while in fact it consists of discrete samples. See also Visualize External Data.

Select Regular Intervals

Aim

To divide your data sets in time intervals of equal length. The results of your analyses will be shown for each interval in separate tables or cells.

You can specify intervals of a specific length (for example, to split your observations in 10-minute units) or a fixed number of intervals. In the latter case, the duration of each interval depends on the observation length and the number of intervals.

Prerequisites

• You carried out your observations.
• The Data Selection window is open (Choose Analyze > Select Data > New/Open Data Profile).

Procedure

1. Click the Settings button on the Results box.
2. Choose either:
   • Use time bins of duration – To specify a fixed interval length. The number of time bins depends on the observation length. If the duration of the observation or event log is not an exact multiple of the interval length, the last time bin will be incomplete. Select the Ignore Last Time Bin if Incomplete option if you do not want to analyze this last interval.
   • Use number of time bins – To specify a fixed number of intervals to analyze. The duration of the time bins depends on the observation length.
3. If your data profile contains two or more result containers (see examples in Create Groups of Data), optionally click Apply to All to set the time bins in all Results boxes.

Notes

• Time bins and the Results box: The Results box shows whether you selected time bins and their duration.
• If an event falls in two or more time bins: If a behavior occurs across two time bins (marked by dotted lines), it is counted in both time bins. For example, its duration may be four minutes in the first time bin and two minutes in the second time bin.
• Maximum number of time bins: In theory you can create as many time bins as you want, but time bins increase calculation time. Make sure you do not have very small, and therefore very many, time bins.
• Time bins and multiple event logs: Time bins start at the observation start time, which is the start of the earliest event log. For event logs that start later, the time bins at the beginning of the observation contain no data.
• The borders of time bins: See What Happens at the Interval Borders?
• Combine intervals with time bins: You can use intervals and time bins simultaneously. The time bins are then generated per interval. The data outside the intervals are not analyzed. For example, if you have two intervals from 0–9 sec and from 20–32 sec, and you define Number of Time Bins = 3, you get three 3-second time bins for the first interval (0–9 sec) and three 4-second time bins for the second interval (20–32 sec).
• The difference between time bins and other intervals: With time bins, you analyze data in regular time segments. Intervals are based on values of events or external data and therefore vary in time. The only exception is by manual selection based on observation time; however, this creates one interval per observation, whereas with time bins you create a number of intervals of the same length within one observation.

Advanced Data Selections

What Do You Want to Do?

• Combine Selection Criteria
• Create Groups of Data

Combine Selection Criteria

Aim

To select data based on a combination of selection criteria.

Prerequisites

• You carried out your observations.
• The Data Selection window is open (Choose Analyze > Select Data > New/Open Data Profile).

Basic Rules

To create advanced data selections, combine selection boxes. Below you find the basic rules to do so.

And Logic

To obtain an And logic, position the boxes after each other in the sequence.

Example: To select intervals by the behavior Playing And filter the behavior Talk to Parent. This way you analyze only the behavior Talking to Parent during the intervals that the subject was playing.

Or Logic

To obtain an Or logic, split the data selection sequence into two branches and place the boxes in parallel.

Example: To analyze those intervals during which a bird either brought a caterpillar or spider into the nest, or a prey smaller than 1 cm.

Do not use this Or logic if you use multiple selection boxes in each branch. The data selection may give unexpected results. Create a separate Results box for each branch instead.

Order of Selection Boxes

The order in which you place boxes in a selection sequence is very important. Please read this section carefully.

If your sequence contains Filter boxes only:

• If each Filter box refers to a different type of element (for example, one Filter box for an independent variable, one for subjects, one for behaviors), then the order you place the boxes does not matter.
• If Filter boxes refer to the same type of element (for example, one Filter (Behaviors) box filtering the behavior group Area of Interest Behaviors, and another Filter (Behaviors) box for Facial Expression Behaviors), then the selection does not work, because the first filter has removed the behaviors in the second filter.