TRAINSET only accepts specific file formats. This sample CSV is formatted correctly. Files uploaded to TRAINSET must have four comma-delimited columms with the header: series, timestamp, value, label; null data are not allowed.
series is a unique name of the time series you are labeling. You can include multiple series in a CSV.timestamp is a timestamp with time zone in ISO8601 format. For example 2019-03-13T21:11:29+00:00 or 2019-03-13T21:11:29Z.value is a numeric scalar. Any real number is a valid for value.label is an integer representation of a boolean; 0==FALSE and 1==TRUE. It is possible to upload data to TRAINSET that is pre-labeled (i.e. the label column does not have to start with all zeros). See the sample CSV.Failure to properly format your data will result in failed uploads.
Labeling is the act of applying labels, whether true or false, to your time series data. Labels can be toggled by clicking on points, or many labels can be applied at once by "brushing" (clicking and dragging) over many points. Navigation of the time series is accomplished either with shortcut keys or by highlighting a "focus region" on the "context bar" underneath the plot. The focus region can be brushed onto the context bar, dragged, and resized with the mouse or with shortcut keys. A legend of available shortcut keys is shown below the plot to help you navigate the time series and apply labels.
After you finish labeling, you can export your data. An export will download a CSV to your default downloads folder. The name of this CSV will have the suffix -labeled. If the file you uploaded already had the suffix -labeled, no additional suffix will be added.
TRAINSET is a client-side application. Although we use the word "upload" to describe importing data into TRAINSET, your time series data and labels are never transferred over the internet or visible to Geocene. Feel free to deploy TRAINSET locally using the instructions in the GitHub ReadMe.