T04 - Project Burma - Singapore Management University

perchmysteriousData Management

Dec 1, 2012 (4 years and 8 months ago)

192 views

T04
-

Project
Burma:

One
-
stop Geospatial
Information of Myanmar

Aung Myint Thein,
mtaung.2008@sis.smu.edu.sg

Htin Kyaw,
kyaw.h.2007@business.smu.edu.sg

Min Thu Han,
thuhan.min.2008@sis.smu.edu.sg

Soe Thet Aung,
soethetaung.2008@sis.smu.edu.sg


Singapore Management Un
iversity, School of Information Systems

Abstract
-

As Myanmar, an ASEAN country, is still an agriculture country, a large portion of
the population still depend heavily on the agriculture industry. Myanmar is surrounded by
Asia's dominant economies such as

China, India, and Thailand. Moreover, the long coastal
line and richness in mineral resources are the pillars of the country's economy. Although it is
still in developing stage, the geographic location gives the country a competitive advantage.
There are
a lot of available data about Myanmar

on the web
.
The lack of
a systematic and
organized presentation of the data
motives us to build this web
-
based application.

Our project
aims to provide users an interactive and comprehensive c
ontent of Myanmar's agriculture
data. Users will be able to get
comprehensive

Geospatial information about the country's
agriculture. Myanmar is div
ided into 17 States.

Index Terms



Geospatial, Myanmar, Agriculture, Rice, Development



1.

M
otivation

Although there are information about
Myanmar and agricultural data, mos
t of
them are available only
on traditional
media storage such as
CDs
. It still lacks a
systematic and organized presentation of
the data on web, which can help users to
visualize these data in a more interactive
way.
Rapid advances in geog
raphic
information systems have created a
potential for dynamic
geo
graphical
visualiz
ation
to be integrated with
explor
atory information visualization
. The
fast growing quantity

of statistical data
accessible
on the Internet creates
po
ssibilities for such integrated
tools to be
used

in a wide range of application
domains. The analysts can
sea
rch for
relationships, patterns
and trends to gain
understanding and knowledge
of
the data,
without having any

prior assumptions or
theories.
As statistics data set
s become
increasingly large and complex, users
require

more efficient
visualization tools

and faster interactive performance. This
challenge demands improv
ed fundamental
methods for
visual explorati
on analysis.
The Web is today’s
prim
ary medium for
dissemination of
geospatial
data and
maps
.
Thus,
our team’s objective is to build an
interactive

web application that can
visualize Myanmar’s social
-
econ
omic data
in a rich, meaningful way

and
provide
more convenient analysis.

The

data

we are using are collected in
statistical surveys by

Food and Agriculture
Organization of the United Nations and
Myanmar Information Management Unit
.
This statistical data

is used mostly by UN
and NGOs

for the purpose of
making

their
policy
d
ecisions, and to facilitate the
appreciation
of economic, social,
demographic
, and other matters of interest
to the
UN
,
various NGOs
,

businesses, and
to the general public. For instance,
population and ec
onomic information is of
great value in
planning
NGOs’ activities

(education, fund allocation,
public
tr
ansport), as w
ell as in private businesses.

The practice of geo
-
referenci
ng census
data has increasingly
s
pread over the last
few decades and the techniques
for
attaching socio
-
economic
data to specific
locations have markedly
improved at the
same

time. In
our dataset
about Myanmar
,
for instance,
most
are provided for each
district.
The data about
these
districts

enable the investigation
of socio
-
ec
onomic
patterns

in association
with the
ir
geographical location. These
cause a
growing demand for more

powerful
exploratory

a
nalysis techniques that can
link population data to their

spatial
distribution. The

web

a
pplication of
integrated information visualization

and
geographical visualization

techniques to
these statistical

data has great potential in
suppo
rting
good policy
decisions and
business decisions
.
Our

statistical
data

is
based on t
he regional divisions in
Myanmar

and is limited
17 states
.

Although the data are available up to 64
districts level, our group decided to
generalize

the data in state level
.
The
chall
enge
is to provide the
users

with
interactive visualization
tools that can
explo
re
the richness
of
statistical
attributes
and help discover
knowledge.


In this
paper, we begin
with a brief review
of
related

work

and references
that has
influenced our

project
, f
ollowed by a
conceptual
and technical description of t
he
system.
Next we

will discuss the
visualization and interaction
tec
hn
iques
implemented. We finish with discussions
with visitors during town hall presentation,

our
c
onclusions
and ideas for future
work
with

this
project.


2.

Related works and
references

Myanmar Information Management
Unit (MIMU):

The MIMU provides a
common information exchange service for
the humanitarian community through
strengthened coordination, collection,
processing, analysis and

dissemination of
information.

The MIMU supports analysis
and decision making by the UN
Resident/Humanitarian Coordinator
(RC/HC), Humanitarian Country Team
(HCT) partners, the UN Country Team and
other actors both inside and outside of
Myanmar.

Digital Ag
ricultural Atlas
:
The digital
agricultural atlas of the Union of Myanmar
is a collection of GIS
-
derived maps,
tabular data and related documents
depicting political, physical and
agricultural resources in Myanmar. Data
are integrated in a “
warehouse”

frame
work, which is the core component
of

FAO's Dynamic Atlas technology
.

The
atlas contains general layers from
international data providers and
agricultural
-
related layers generated from
2001
-
2002 statistics

collected at
state/division, district and township level.
Tabular information and maps are linked
to build a geographical information system
for agricultural resource managers.

Protovis
: Protovis composes custom
views of data with simple marks such
as
bar
s

and
dots
. Protovis
provides
dynamic
properties that encode data, allowing
inheritance, scales
and
layouts

to simplify
construction
.

Protovis is fr
ee, open
-
source
and uses JavaS
cript for web
-
native
visualization, which consists

a great deal
of our project
.


3.

Methods and s
ystem
implementation

Throughout our project, we went through
several iterations


we test our application,
we fo
und problems and we fixed them.

Iteration
ONE
:

Firstly, our team extracted
the data from Myanmar Information
Management Unit and

Digital Agriculture
Atlas Database. PostgreSQL
was used
to
read the data shape files. Next, we used
PHP as
a
medium to read from
PostgreSQL and convert the data into
GeoJSON so that Protovis and Polymap
can process those data. Polymap library

was used

to
draw our map and to present
the data map.
Since g
raphs are the best
tools to present the data in int
elligent and
analytic sense and
our project objective is
to present the data in a useful and insight
intelligent way, we
chose

Protovis library
to draw the
graphs and to present our
agricu
ltural data.
In order to

deliver a
better

data visualization application, we
tried to link the map and the graphs. Our
objective is to present the data in both map
and the graphs
for
an
y

chosen
attribute
of
the agricultural
data

(
such as

rice,
populatio
n, mechanization or chemicals).

Problem
: Although we were able to
link

the map and the histogram,

the map and
the histogram were not updated
spontaneously when we choose another
attri
bute
.
The problem was spotted.
Although Poly
map responded

to dynamic
data changes from php and re
-
r
ender itself,
the Protovis does

not have built
-
in re
-
render function to follow the nature our
dynamic data changes.

Iteration TWO
:
As refresh
ing

the
histograms when a user selects different
attribute

i
s needed
, we tried to tackle that
problem in this iterati
on. We used iframe
approach


we
contained the histogram
inside the iframe and forced refresh when
a different attribute is chosen
.

Problem
: The histograms could

be
refreshed because of i
frame. Howev
er, the
histogram was

isolated from the whole
application, which means that the
histogram and the map drawing
were able
to

respond

properly.
Thus, our team
looked at

other options. We were still
using PostgreSQL as a component inside
the main application i
n this second
iteration.

Iteration THREE
:

Our team figured out
a way to refresh our histogram inside the
page without using iframe.
Firstly, w
e had
to clear the panels with drawing, and use
self
-
created function to re
-
render the
graphs. Moreover, we
gener
alized

the map
boundaries
in order
to improve the
performance of our application.
Protovis
was used for both map and graphs instead
of using PolyMap for map.

Another
important change for this final iteration is
that we remove the PostgreSQL from our
application layer. As
the
following system
architecture diagram

indicates
,
PostgreSQL

is used

only when the data
were extracted
from the
initial
shape files.
Then,

the shape files
are imported
into
Postg
reSQL and

php
is used
to convert the
data to the desired data format for graphs.
Because of the flexibility of the php, the
data
are convertible
into any array format
required for the Protovis graph libraries.
W
e use
d jQuery for the user interface.


Figure
1
: System Architecture


4.

Data Visualization
Techniques


4.1

Application Interface


The main page is shown above, consisting
of
-

1.

categories of data on the left, which
users can select in order

to view

2.

Thematic map showing the
distribution of data based on
geographical locations

3.


Three different kinds of data
graphs(Histogram, scatter plot and
box plot) on the right

4.

Description of the categories
selected

T
he distribution of rice
, the use of
mechanization and the use of chemicals
among the 17 state areas

are selected as
data examples for our applications.
Moreover,
demographic data such as
population is also chosen. The followings
are the categories of data currently
presented in our applicati
on:

1.

Population (M, H and B)

1.

Distribution of total
population

2.

Distribution of male
population

3.

Distribution of female
population

2.

Rice (M, H, S and B)

1.

Distribution of Sown Area
in Acre

2.

Distribution of Harvest
Area in Acre

3.

Distribution of Yield in Kg

4.

Distribution of Productivity
in Tonne

3.

Mechanization (M, H, S and B)

1.

Distribution of use of
Power tillers

2.

Distribution of use of
Harvesters

3.

Distribution of use of
Reapers

4.

Chemicals (M, H, S and B)

1.

Distribution of use of Urea
in Tonne

2.

Distribution of use of Triple
superphosphate in Tonne

3.

Distribution of use of
Muriate of potash in Tonne

4.

Distribution of use of other
fertilizers

in Tonne

5.

Distribution of use of
Pesticides in Tonne

(M = Map, H = Histogram, S = Scatter
plot, B = Box plot
)

4.2.
Thematic Map

The map is the best tool to present the
geospatial data how they are distributed
and how they are related to each other. For
example,
just by
look
ing

at the data of the
distribution of yield of rice around
Myanmar in numbers,
the user

cannot
make much useful information from it
without presenting them on the map. All of
the data shown in above are spatially
related to each other among 17 sta
tes.
Therefore, when
those data
are presented
on the map, the users
can get more insight
and int
elligent information of how
the data

are spatially distributed and how they are
linked and related to each other. The map
boundaries are simplified to improve the
performance of the application.
Our team

used 17 officially

divided states so that the

data
a
re represented
in more details.

Moreover, Choropleth
map
method
is used
to present
the map. Collected data are
aggregated by state and
colo
u
r
ed

according to the percentile they are in.

4.3.
Histogram


H
istog
ram can help
user
s

see how the data
is distributed among 17 states in a short
glance.
Users are provided with

the
options to choose the number of bins, 5, 10
-

default and 20. The example screenshot
is the distribution of the sown area of rice
in acre.
T
here is a state whic
h has a very
large sown area for rice. Although
the user
might know that
rice is the major crop in
Myanmar,
he can clearly see from the
histogram that 8 states

are under the first
range (minimum).
The users can adjust the
level of details by choosing diffe
rent
numbers of bins
.
Moreover, as graphs and
map are linked and interactive, users

can
click on the bars from the histogram and
check which states are in the first range or
which state has the largest sown area of
rice on the map.



4.4.
2D Scatter Plot




Scatter plot can present the data in three
parameters in 2 axis graph.
T
he
distribution and the relation of the data

can
be observed

in a very quick and effective
way. However, it is important to consider
which parameters will be on which axis to
present t
he data in a meaningful way. For
example, the
above
screenshot provided is
the scatter plot for sown area of rice (y
-
axis), the number of power tillers used (x
-
axis) and the productivity of rice in tonne
(z
-
axis or the area of the cycle).
The user
might ha
ve already known that this
particular

state has a very large sown area
of rice compared with other states.
However,
he

cannot extract much
information from it

alone
. W
ith the help of
scatter plot, the user

can see that state has
not only very large sown
ar
ea;

it also has
very large number of power tillers used
and very large productivity too.

4.5.
Box Plot


Box plo
t helps a user visualize
the
distribution of the data
as a whole
in a very
quick way. Because it has only one thing
(box) to look at, user won't be able to miss
out the information that a box plot want to
present. The example screenshot is the box
plot of the sown area of rice.
The figure
indicates that
sown are
as of ri
ce in
Myanmar are

not equally distributed. 75%
of the state has sown area lower than the
medium sown area while the highest 25%
of states (only one state according to the
histogram) has very larger sown area.
Therefore,
the user can observe
the
distributio
n of sown area of rice in a whole
scale in a very short glance.



5.

Discussion

Our application is intended to give the
users a comprehensive visualization of the
demographic and agricultural industry of
Myanmar. In other words, visualization
offers
integrated exploration of
geographical visualization using thematic
map and statistical visualization using
various Visualization and analysis of
agricultural industry’s data based on
geographical locations will help user to
fully grasp understanding of th
e industry
and supports decision making. During the
town hall presentation, visitors were
initially attracted by the word “Burma”
and some of the initial user responses are
“How do you get these data?” “What is the
unique selling proposition of your
applic
ation?


A web
-
enable application
that presents agricultural data with
Myanmar map and various charts makes
the visitors interested in the application.

Most of the user expected the short
introduction of what the application can
do. We walked them through
from the
initial idea of the application to the details
of some application features. Explanations
were on how the industry data were
presented on the map so that the users can
know exactly which part of the country
represents specific data even if they do
n’t
have knowledge of the region names. For
instance, the users can see which part of
the state has highest population by looking
at the histogram and when the respective
bar in histogram is clicked, the
corresponding area on the map is
highlighted. After

our walk through of the
application, some of the users explored the
visualization themselves. Some of the
visitors were buffled by the use of
box
plot
, which they found difficult to
understand. One visitor asked us to go into
details of the scatter plot o
f rice
productivity. He question was regarding to
the use of data for X/Y axis and their
relationship. He pointed the application
would be more meaningful if it lets user to
change dataset for X/Y axis. More
questions were made on the data source,
and the
programming language used for
the application. Our team was glad to
receive valuable suggestions,
recommendations from the people in the
GIS industry during town hall presentation
user assessment. First of area for
improvement defined by users was to
enabl
e user to click on the scatter plot data
and which, in turns, would highlight the
respective area on the map. Another was to
let user upload new data and let
application present the data in the map and
graphs. We take in some of the idea and
improved our f
inal visualization with
highlight features of relevant labels on
scatter plot, connection between the scatter
plot’s data and the map, and improved
presentation of the graphs and use of
colours
. Another important feedback was
that the map’s boundary is a
bit
generalized. However, we explained that it
was to make the application smoother.

In conclusion, our application needed
improvement on the data analysis part but
the visitors are enlightened by presentation
of data in terms of colourful thematic map
and various graphical representations,
which are explanatory every user.


6.

Future Development


Based on t
he experiences and techniques
learned

f
rom this application project,
our
application
not only can be improved
but
also can
be
develop more advanced and
flexible including but not limited to the
following solutions.

Trade area analysis application

We can develop an application to process
the trade area analysis for business
-
users.
One of the examples is to develop an
application to
analyse

the factory site.
Assume that state A has more expensive
land price than state B but state A is the
target state. If we want to build a factory in
state B, we need the transportation from
state B to state A. However, if we consider
about the transportatio
n of raw materials,
the storage of raw materials, and other
hidden costs in state A, it might be
possible that state B is preferred than state
A. All of
these

analyses

can b
e done with
Trade Area Analysis
provided that we
have enough data.

Service Provide
r

As our application doesn't have special
data format and we simply consume text
array as data, we can open up the service
to the customers not limited to Myanmar
but to all the other countries. One simple
development is let user upload the data
such as po
pulation, rice distribution of
another country and we can provide them
the same analysis that we have done to
Myanmar data. The more we can
generalize our application, the more
scalable

our application will become to
consume customers' data and it will
bec
ome the better.

Secondly, as we have now figured out the
way to forced re
-
rendering of Protovis
graphs and maps, we can look back to our
Iteration 1 architecture where live Data
Layer (PostGIS and PHP) and Application
Layer (Protovis + Javascript + Ajax +

JQuery ) connection was considered for
scalability and easier data format. If we
can achieve this
architecture,

our service
can accept standardized data format such
as
Shape file

with attribute columns to be
directly uploaded by user instead of
proprietar
y format. The architecture will
still able to accept the simple excel and
text file

with appropriate district
(division

)
code column and attribute column.

7.

Conclusion

This application allows

user to visualize

geographical statist
ical data about
Myanmar. By allowing users to view data
not only on thematic maps, but also on
histogram,
scatter plot

and
box plot
, the
application makes the data visualization
easier and convenient. With the future
development added in,
this
a
pplication
of
integrated information visualization and
geographical visualization

techniques to
these statistical data has great potential in
suppo
rting good policy decisions and
business decisions for NGOs and foreign
investors who need an initial information
resear
ch on the Myanmar.

8.

Acknowledgement

Our team would like to thank Professor
Kam Tin Seong for his guidance and
consultation sessions, our classmates and
people who came down to our town hall
presentation for their valuable feedbacks.



Bibliography

1.

Tailor
-
made Exploratory Visualization for Statistics Sweden . Retrieved on April 2011 from
http://servus.itn.liu.se/projects/geowizard/documents/feldtn_statisticssweden.pdf

2.

Digital Agricultural Atlas. Retrieved on April 2011 from

http://dwms.fao.org/atlases/myanmar/overview_en.
htm

3.

Myanmar Information Management Unit (MIMU). Retrieved on April 2011 from
http://themimu.info/

4.

JavaScript tutorial. Retrieved from
http://www.w3schools.com/js/default.asp

5.

PHP official website. Retrieved from
http://php.net/

6.

Protovis Library. Retrieved from
http://vis.stanford.edu/protovis/