{timeseriesdb}: A Time Series Database for Official Statistics

-> GitHub Pages Documentation Site <-

{timeseriesdb} maps R time series objects to PostgreSQL database relations for permanent storage. Instead of writing time series to spreadsheet files or .RData files on disk, {timeseriesdb} uses a set of PostgreSQL relations which allows to store data alongside extensive, multi-lingual meta information in context aware fashion. {timeseriesdb} was designed with official statistics in mind: It can keep track of various versions of the same time series to handle data revisions, e.g., in the case of GDP data.

Why {timeseriesdb} ?

{timeseriesdb} ...

is lite weight but powerful: multi-language meta information, versioning of time series, ...
built entirely based on license cost free open source components.
tailored to the needs of Official and Economic Statistics
administration friendly, extendable access rights management
well documented, developer friendly.
API ready: {timeseriesdb} can easily be extended to expose data through a REST API to allow for language agnostic access to your time series.

What Does {timeseriesdb} NOT DO ?

{timeseriesdb} is not built to incrementally append new observations as fast as possible. {timeseriesdb} does not try to compete with the amazing speed of InfluxDB. It's not a server log or IoT minded time series storage.

Quick Start Guide

Make sure you followed the installation notes to make sure all components of the {timeseriesdb} were installed properly: PostgreSQL, necessary PostgreSQL extension, R as well as the {timeseriesdb} R package.

Example Use (Basic Usage)

The following examples illustrate basic use in a nutshell. The learn more about the use of {timeseriesdb}, read the vignette articles.

Store a List of R Time Series Objects to the Database

# Create DB connection. 
# In this case connect to a local db running on port 1111
# /w lame passwords -- strongly discouraged for production. 
con <- dbConnect(Postgres(),
                "dev_writer", "localhost",
                 1111, "dev_writer",
                "postgres")
tsl <- list(ts1 = ts(rnorm(100), frequency = 12,
                     start = 2002),
            ts2 = ts(rnorm(100), frequency = 12,
                     start = 2001))
db_ts_store(connection, tsl)
dbDisconnect(con)

Read Data into a list of R time Series object

con <- db_connection_create(
  dbname = "postgres",
  user = "dev_admin",
  host = "localhost",
  passwd = "dev_admin",
  port = 1111
)

tsl <- db_read_ts(connection, c("some_ts_id","another_ts_id"))
db_connection_close(con)

Advanced Features

{timeseriesdb} offers a plethora of features beyond just mere storage of time series themselves:

store vintages (versions) of time series
datasets to group time series
store extensive, multi-lingual, versioned meta information at dataset and time series level
individual, user specific collections of time series similar to bookmark or playlist functionality
administration friendly access management with reasonable defaults (public, internal, restricted)
release calendar functionality to facilitate publishing

mbannert / timeseriesdb

readme