Upload a normalized database

In this article, we explain how to prepare a data file for upload to Knoema by creating a three dimensional database in Microsoft Excel.

In the very first row, the file is expected to have dimension names. First of all, start addding dimensions which are specific to your data. Number of dimension depends upon a database you want to upload. For example, for an agricultural outlook such dimensions could be: Country, Commodity and Indicator.

Knoema has some predefined dimensions which are common for all data, they are the following:

  • Date. The Date is a required field. It defines a calendar dimension of your data. it stores dates of various data points.
  • Frequency. Indicates a frequency of a time series. Knoema supports the followind frequencies:
    • Annual (A).
    • Semi-annual (H).
    • Quarterly (Q).
    • Monthly (M).
    • Daily (D).

Knoema also predefines Value and Unit fields. Value will include actual figures, Unit will host units of measurement associated with a data value such as $US, Metric Tonnes, Hectars, etc. As a result, the first row of the file might look like the following:

 

Now, start populating columns with names or codes of dimension elements. Knoema will use these values when showing you the data in the Dataset Viewer or any other Knoema tools. For our demo case, elements could be:

  • Country: Argentina, Australia, Austria
  • Commodity: Rice, Soybean, Wheat
  • Indicator: Area Harvested, Production, Domestic Consumption, Import

Each unique combination of those dimension elements together with Frequency field will be treated as a time series.

When providing values for Date field, consider the Frequency of a time series. Use the very first day of a period. For example April 1 is the first day of Second Quarter of a year, February 1 is the very first date of February monthly data, July 1 is the first day of second semester, and so on. Use specific dates when uploading daily data.

Below is an example of the file ready for upload to Knoema.

Note that Data andFrequency fields could be merged. In that case, Date field is expected to have dates provided in statistical notation and Frequency column is not required. See below some examples of dates following statistical date format:

  • 2009, 2010, 2011 - Years
  • 2009H1, 2013H2 - Half-years
  • 2009Q1, 2010Q3, 2012Q4 - Quarters
  • 2009M2, 2011M7 - Months
  • MM/DD/YYYY, DD.MM.YYYY - Days

Feedback and Knowledge Base