Statistics Canada workflow example

19 March 2025 - Written by ML

1. Introduction

In this report we present a workflow for pulling Canadian statistics and demonstrate how to extract data on employment in the oil and gas extraction industry. The approach uses Statistics Canada’s bulk delta daily data file1 along with data from the Statistics Canada website. This combined approach helped to understand the data file structure and rapidly identify the specific vector code2 for our series of interest

2. Workflow Overview

Statistics Canada produces a daily file called delta that contains statistics data. When downloaded as a ZIP file, the delta3 package includes the data, metadata, and a code file. Content within the code file can be used to map data onto both meta and data file fields. And the meta file is connected to the data file through a common field - productId which is the numeric table identifier.

After unzipping the container, the metadata and code files (which are in XML format) are converted into CSV format.

The Statistics Canada website is then used to identify the series of interest. And the vector ID is found from within the created database file that contains the specific series data.

3. Data File Structure

The data file includes several key fields such as:

  • productId: Reference for the table of interest.
  • vectorId: Unique series reference.
  • refPer: Reference period for each series value.
  • value: The statistical value.
  • coordinate: Describes the dimensions for all series available within the table
  • other fields show in the table
Overall Data Dimensions
Metric Value
Number of Rows 61847371
Number of Columns 13
Memory Usage 3.9 Gb
Data File Structure, Types, and Sample Values
Column Data_Type Sample_Value
productId productId integer 10100139
coordinate coordinate character 1.12.0.0.0.0.0.0.0.0
vectorId vectorId integer 39050
refPer refPer character 2025-03-12
refPer2 refPer2 logical NA
symbolCode symbolCode integer 0
statusCode statusCode integer 0
securityLevelCode securityLevelCode integer 0
value value numeric 2.9991
releaseTime releaseTime character 2025-03-17T08:30
scalarFactorCode scalarFactorCode integer 0
decimals decimals integer 4
frequencyCode frequencyCode integer 1

4. Metadata and Linking

The metadata file provides supplementary information about all series related to the table of interest. It includes details on the dimensions (e.g., geography, industry classification, release date) that can be used to filter and link data. The key linking field is productId, which connects the data file with the meta file.

The key meta file field of interest is DimensionsSummary that provides an understanding to the coordinate field in the data file. In the case of the employment table there are four dimensions listed including: Geography (Pos 1): 1 members; Estimate (Pos 2): 2 members; North American Industry Classification System (NAICS) (Pos 3): 435 members; Release (Pos 4): 120 members.

By using the website for this particular series these categories are clearly shown. For our example we have selected:

  • Pos. 1 - Canada;
  • Pos. 2 - Employment in units;
  • Pos. 3 - Oil and Gas extraction industry [211,2111] out of a total of 435 industry series; and,
  • Pos. 4 - latest release 2025-02-27 out of a total of 120 releases.

Once the criteria were selected in the website the data could be downloaded. Of the various download file options the one that contained the data in database format contained the vector code which is the unique identifier for our series of interest.

After obtaining the vector code we could then extract the precise series with its reference date and value and place that into an R dataframe for further analysis and visualization.

Overall Meta Data Dimensions
Metric Value
Number of Rows 1
Number of Columns 21
Memory Usage 29.7 Kb
Meta Data File Structure, Types, and Sample Values
Column Data_Type
ProductID ProductID integer
CansimId CansimId character
CubeTitleEn CubeTitleEn character
CubeTitleFr CubeTitleFr character
CorrectionID CorrectionID logical
CorrectionDate CorrectionDate logical
CorrectionNoteEn CorrectionNoteEn logical
CorrectionNoteFr CorrectionNoteFr logical
CubeStartDate CubeStartDate IDate, Date
CubeEndDate CubeEndDate IDate, Date
NbSeriesCube NbSeriesCube integer
NbDatapointsCube NbDatapointsCube integer
ArchiveStatusCode ArchiveStatusCode integer
SubjectCode SubjectCode logical
SurveyCode SurveyCode logical
frequencyEn frequencyEn character
FrequencyFr FrequencyFr logical
responseStatusCode responseStatusCode integer
DimensionsSummary DimensionsSummary character
FootnotesEn FootnotesEn character
FootnotesFr FootnotesFr character

5. Next step

The next step in this approach is to find an efficient workflow for rapid identification of all unique vectors for each table. And then to establish a work flow that would allow iteration over all dimension series for certain statistics. For example if we wanted to understand variance in employment for all employment sectors across each month per year.

References


  1. [as of 2025-03-17]↩︎

  2. [1645360662]↩︎

  3. [url = “https://www150.statcan.gc.ca/n1/delta/20250312.zip”]↩︎