Video summary
The video demonstrates how to integrate external population data from WorldPop into the DHIS2 system using its Maps application to facilitate comparative analysis. By adding a new population layer, users can overlay global estimates onto their specific country boundaries, such as the example provided for Laos. This process allows the system to automatically calculate population figures based on available polygons and years, though it is noted that data availability may vary depending on the selected geographic boundaries. Once the external layer is loaded from sources like Google Earth Engine, users can access a data table to view province-specific estimates, which serve as an external benchmark for comparison against internal census data stored within the organization's system.
To visualize these comparisons effectively, the presenter guides viewers through creating a new dashboard that juxtaposes internal census projections with external estimates for a specific year, such as 2020. In this setup, the internal data represents local statistics bureau figures broken down by age and gender, while the external tab displays total population estimates derived from WorldPop. The visualization is configured to filter by province, allowing analysts to see side-by-side values for each region. This approach provides a practical starting point for understanding discrepancies between different data sources without requiring complex manual imports or spreadsheet manipulations, although those methods remain valid alternatives for handling multi-year projection data.
The analysis reveals significant variations between the two datasets, highlighting the importance of evaluating external sources before relying on them for decision-making. For instance, in Vientiane Capital, the internal estimate is approximately 885,000 while the WorldPop figure is around 967,000, showing a moderate difference. However, in Phongsaly province, the discrepancy is stark, with the external source estimating only 40,000 people compared to the internal census data of 187,000. Such large gaps suggest that WorldPop might not be accurate for certain regions or contexts, indicating that it may not always be a suitable primary source without further validation.
Ultimately, the video concludes that while external tools like WorldPop offer a convenient way to quickly populate missing data or provide baseline comparisons, analysts must critically assess the accuracy of these figures against their own verified census data. When significant differences arise, as seen in the Phongsaly example, it is advisable to investigate potential reasons for the error or explore alternative platforms such as UN population projections. By systematically comparing multiple sources, organizations can better understand the limitations of global datasets and make informed decisions about which external estimates are reliable enough to supplement their internal records.
Read the full video transcript
We do have in DHIS2 um
population data from uh WorldPop, okay?
And this is accessible via the maps app.
So, I'm going to go to the maps app,
okay?
And I'm going to add a layer,
and I'm going to add this population
layer cuz it's going to take the
uh structure, you know, of your country.
It's going to take the the polygons uh
as boundaries um for your country and
calculate this population. This
population is taken from a different
source than than what's in your system,
I would imagine, okay? So, when I add
this new population layer, um
I have uh some data here,
okay? Uh I'll just keep this uh
uh by uh by default as as what I uh see
here. I have a period, and what it's
going to do is there's not data
available for every year
based on the boundaries I've selected,
and in this case I'm using Lao as my
boundary. You know, your own system will
have a different country boundary, of
course. Um it's going to show you the
latest available uh periods that are
available um for for the country based
on your boundaries, okay? In this case,
I only have data available for 2020, so
I can use this as a source of external
comparison, okay?
I'll then select my organization units,
and what I want to do is I want to get
all the provincial data um for my
country. That way I can compare cuz I
will have provincial um census data for
2020, okay? Um for my population in my
system, and I can compare it with the
provincial data that is used here in
WorldPop, okay?
And then I'll just leave the style just
as it is, okay?
And I'll add this layer, okay? And it's
going to take a second
to get this data from uh WorldPop and
Google Earth Engine,
okay? And it's going to create a map.
Now, okay, the map's nice, but that's
not really what I'm after. Okay?
When I go here to more actions, I can
click on this option that says show data
table. Okay? And if I open that up, I
now get
in a tabular format
the province
and the population estimate
for that province. Okay? From an
external source.
In this case, it's WorldPop, and in this
case the year is 2020. So, I can create
a table now
in DHIS2 using my population data, and I
can compare it with this data just as a
start.
Right? Just as a start. So, I mean
in that first dashboard that I had, I
had many more years of data um
from the UN population projections,
right? But you would need a little help
maybe to import that data into your
system. You could also put it on a
spreadsheet, right? Uh that's a very
valid method. You could take your data
from DHIS2, download it, and then put in
for various years um
data for from the these population
projections in a spreadsheet. Okay? So,
there's there's a lot of ways we can
kind of look at this. Okay? But now that
I have this data, what I'm going to do
is just go to data visualizer here, and
I'll make a new visualization.
Okay? And I'm just going to make that
the same visualization essentially. So,
in this system,
we have this Let's see. Is there a
group?
Filter.
No. Okay. So, I'm just going to search
for it. It's called LSB. It stands for
the Lao Statistics Bureau. Okay?
Um so, this is where our population data
is stored for this demo system. Um, you
know,
if you're using your own system,
obviously that this will be a different
place, okay? And what I want is the
total population by province for 2020,
okay? So I can compare it with the data
that's there. This is just as an
example, just as a start to give you
some experience in uh, you know,
analyzing these differences. Okay?
So I will have a look here.
My population is stored as a single age
and breakdown by male and female here,
but I'm going to get the total, so this
LSB population estimated single age,
this is going to show the total
population for me, okay?
Um, so I'll hide this and then what I
want is just for 2020,
okay? So I can select a fixed period
and I'll select 2020 in this case,
okay? And I want the provinces in this
case cuz that's the same data that I've
obtained from my external source.
And I just want to
change my layout here.
Can filter out.
Okay, so now I have
estimated populations based on my own
projections, right? This is from my
census
and in this other tab
I have estimated projections based on an
external source,
right? And I can start comparing these,
right?
So here for Vientiane Capital, I have a
value of 967,000
give or take, right?
And my value is 885,000,
okay?
So there is a difference there,
whether we want to mark it as
significant or not, it might be worth
looking at some of the other ones
and just seeing
what the case is here. Here we have a
much more stark difference.
Phongsaly, the second province,
okay?
We have a value of 40,000
for our external source,
but in DHIS2, our data is 187,000.
Now, the reasoning behind this
difference could be quite varied.
That WorldPop might not be very accurate
in this case, right? So, we might not
really
want to target it as a source, right? We
could then look at other platforms,
the UN population projections or other
types of population projections,
um and also see if there's some
variation there, right? As well, and
start comparing them, right? To what we
have here.