Data import into DHIS2

Jason Pickering

August 6, 2017

Set your login credentials and server

You may need to change the following to suit your environment.
Change the string in setwd to match the location of the GitHub repo on your machine.
Change baseurl to point to where your installation of DHIS2 is running.
Change username and district to suit your DHIS2 installation.

setwd("/home/jason/development/dhis2-data-munging/")
baseurl<-"http://localhost:8080/dhis/"
username<-"admin"
password<-"district"

Load some libraries and login to DHIS2

## [1] TRUE

Loading the data

Get the data from the UNAIDS website
We will assemble a request for the http://aidsinfo.unaids.org/ site and parse this data into a data frame.

Loading the data

We have to spend a bit of time to look at the response format, but it seems to be a JSON object which we will need to reconstruct into a flat table.
Create two tables for the year and country
Unnest the data contained within the array.
There are better ways to do this. :)

All ages estimate	All ages lower estimate	All ages upper estimate	country	year
369.55116	142.92316	745.52936	Afghanistan	1990
483.6869	205.20409	965.83533	Afghanistan	1991
622.1754	264.81107	1203.39917	Afghanistan	1992
778.90703	357.96414	1485.74292	Afghanistan	1993
944.91716	449.15015	1733.13586	Afghanistan	1994
1105.99481	555.44891	2007.5	Afghanistan	1995

More data cleanup

Lets only get the middle estimates, ignoring the lower and upper estimates.
Lets round these numbers to the nearest integer.
Remove any “NA” or NULL values. They will not be imported.

d<-d_all[,c("All ages estimate","country","year")]
names(d)<-c("value","orgUnit","period")
d<-d[!is.na(d$value),]
d$value<-round(as.numeric(d$value))

## Warning: NAs introduced by coercion

knitr::kable(head(d))

value	orgUnit	period
370	Afghanistan	1990
484	Afghanistan	1991
622	Afghanistan	1992
779	Afghanistan	1993
945	Afghanistan	1994
1106	Afghanistan	1995

Matching the orgunits from DHIS2

Get a listing of countries from DHIS2. We will need to match these with our data.
Convert the country names in the data file back to a character (instead of a factor) so that we can use this column to join the OU listing from DHIS2 with it.
Very likely, this is not going to match, so let’s try to first perform a LEFT JOIN on the data.
This will keep all of the countries, even if there is not a match from DHIS2.

## [1] "Bolivia (Plurinational State of)"     
## [2] "Cabo Verde"                           
## [3] "Côte d'Ivoire"                        
## [4] "Democratic People's Republic of Korea"
## [5] "Myanmar"                              
## [6] "Republic of Korea"                    
## [7] "Russian Federation"                   
## [8] "South Sudan"                          
## [9] "Venezuela (Bolivarian Republic of)"

So, we see we have a number of countries which do not match.
We can define a mapping to convert the names in the data file to the names in DHIS2.
We are still going to be missing a few countries (like South Sudan ) but our hierarchy does not have this country, so we will just have to ignore it.

from	to
Bolivia (Plurinational State of)	Bolivia
Cabo Verde	Cape Verde
Côte d’Ivoire	Cote d’Ivoire
Democratic People’s Republic of Korea	Korea, Democratic People’s Republic of
Myanmar	Burma
Republic of Korea	Korea, Republic of
Russian Federation	Russia
South Sudan	South Sudan
Venezuela (Bolivarian Republic of)	Venezuela

The mapping looks OK, so lets reassign the names.
This time we will perform an INNER JOIN and discard any data which does not have a matching country in DHIS2.

Create a new data element for the data

We will need to create a new data element for the data in the system.

## Warning in set.seed(38894): '.Random.seed' is not an integer vector but of
## type 'NULL', so ignored

We get the data prepared to import into DHIS2.
Set all of the property names according to those used in the documentation.
Upload to DHIS2 and hope that it works.
Finally, trigger analytics.