Time-Series Data : Preparation

This page is part of Exploring Time-Series Data.  Examples use the world development indicators dataset.


Time-Series Data Input Format

To describe your time-series data, create a separate column for each time-key (such as each year). Each column name should be written as  AttributeName->TimeKey. The attribute name shows what is being measured (for example, GDP), and time key is when the data was recorded (a year, or year-month, etc.). The screenshot below shows  a sample dataset with timeseries attribute "GDP PerCapita" by "year".


Detecting time-series using the derive menu

When you import a dataset with time-series, Keshif will first recognize each time-key (such as year) separately, and present them as individual numeric attributes. To prepare the time-series data, click "derive" on any of these numeric attributes, and select the key type among the options. A new "time-series" attribute will be added, the individual numeric attributes will be hidden, and most important of all, the analytics options for time-series data will be enabled.

Notes: 

  • You can also specify a custom format using the the    https://github.com/d3/d3-time-format.
  • We will be releasing scripts to help you prepare your time-series data from some other common data representations, such as one row for each time-key (such as year). Contact us if you need support.

Detecting time-series in dashboards using the Boost features

Keshif's "boost" feature can detect time-series data where the time-keys are years or numeric values.  This is also a great way to prepare many time-series attributes at one step! 

In the boost results, you can review all the suggested changes, and disable/enable them individually.

Notes:

  • The boost currently detects time-series attributes indexed by year/order, and it does not yet recognize other time formats. You can use the derive menu to describe other time-series formats.
  • The data within all the time-keys should be all numeric. Please review your data for columns that might include categorical/text data.  Specifically, formula errors in Google Sheet are converted to text values. You can avoid text values that denote errors by wrapping up your Sheet formula inside IFERROR( ... ) to remove the "error" text from the cell. See  https://support.google.com/docs/answer/3093304 for details.

Still need help? Contact Us Contact Us