The visualization highlights the relative strengths of political parties and coalitions that won 20 or more seats in the 2011 National Constituent Assembly elections.  Each district has four percentages reflected: green represents CPR (which won 29 seats), gold represents Aridha (26), orange represents Ettakatol (20), and blue represents Joumhouri (20); the percentages for Joumhouri reflect combined levels of voter support for  PDP (16) and Afek Tounes (4) which merged after the elections.  

The textual heat map of political party results relies on two data sets:

• Tabular data for the four parties and coalitions that, in addition to Ennahdha (89 seats) won 20 or more seats in the 2011 elections.
• geometry for the 27 governorates we will derive from the delegation shapefiles.

1. Dissolving Delegations into Districts

Since the party data has a district-level resolution, we need to create district polygons out of delegation polygons that we already imported.

To do this you’ll need to load the filter_rings() function explained in this post. This functions removes artifacts left by unioning each district’s member delegations. Once you’ve loaded the function into your database, run the following query.

create table districts as (
select
	circo_id,
	filter_rings(st_union(the_geom),2) as the_geom
from delegations
group by circo_id
);

2. Importing party data to PostGIS

If you have not already, you will need to import the delegation shapefiles. Detailed information for importing shapefiles can be found in the Data management + mapping section of Setup.md.

To import the tabular data, first set up 4 tables - one for each party (Aritha, CPR, and Takattoul, Joumhouri). Run the following sql in the psql shell.

create table takattoul (
	district_id int,
	district text,
	active_votes numeric,
	active_ratio numeric,
	automatic_votes numeric,
	automatic_ratio numeric,
	total_votes numeric,
	total_votes_ratio numeric
);

create table Aritha (
	district_id int,
	district text,
	active_votes numeric,
	active_ratio numeric,
	automatic_votes numeric,
	automatic_ratio numeric,
	total_votes numeric,
	total_votes_ratio numeric
);

create table cpr (
	district_id int,
	district text,
	active_votes numeric,
	active_ratio numeric,
	automatic_votes numeric,
	automatic_ratio numeric,
	total_votes numeric,
	total_votes_ratio numeric
);
create table Joumhouri (
	district_id int,
	district text,
	active_votes numeric,
	active_ratio numeric,
	automatic_votes numeric,
	automatic_ratio numeric,
	total_votes numeric,
	total_votes_ratio numeric
);

To populate the tables you just set up, run these commands from your command shell. They are labeled for your reference.

# Aritha
cat 2011data/resultsPerParty/Aritha.csv | psql -d tunisia -c "COPY Aritha FROM STDIN WITH DELIMITER ';' CSV HEADER"


# takattoul
cat 2011data/resultsPerParty/takattoul.csv | psql -d tunisia -c "COPY takattoul FROM STDIN WITH DELIMITER ';' CSV HEADER"

# CPR
cat 2011data/resultsPerParty/cpr.csv | psql -d tunisia -c "COPY cpr FROM STDIN WITH DELIMITER ';' CSV HEADER"

3. Joining all three parties’ tables

In order to see data for all three parties associated in one table, we will need to join two parties’ tables together for joining the third. Thankfully the district_id key is consistent across tables. To join, run

select
	b.ratio_t,
	b.ratio_c,
	n.total_votes_ratio as ratio_n,
	n.district_id,
	n.district
from cpr n join 
	( select 
		  t.total_votes_ratio as ratio_t,
		  c.total_votes_ratio as ratio_c,
		  c.district_id
	    from takattoul t join cpr c
	  on t.district_id = c.district_id
	) as b
on n.district_id = b.district_id;

In addition to joining the relevant columns into one table, this query aliases the total_votes_ratio column from each table as ratio_x where x is the first letter of each party’s (and table’s) name. It will be important to remember the naming conventions implemented now, as they will have implications at various steps downstream.

4. Joining spatial data to tabular data

To join the tabular data compiled in the previous to the district polygons we created earlier, run the following SQL:

create table parties as (
	select
		g.circo_id,
		g.the_geom,
		d.district_id,
		d.district,
		d.ratio_t,
		d.ratio_n,
		d.ratio_c,
		coalesce(cast(round(d.ratio_t) as text)||'%', 'n/a') as ratio_tc,
		coalesce(cast(round(d.ratio_n) as text)||'%', 'n/a') as ratio_nc,
		coalesce(cast(round(d.ratio_c) as text)||'%', 'n/a') as ratio_cc
	from districts g left join 
	  ( select
		  b.ratio_t,
		  b.ratio_c,
		  n.total_votes_ratio as ratio_n,
		  n.district_id,
		  n.district
	  	from nahdha n join 
		  ( select 
			  t.total_votes_ratio as ratio_t,
			  c.total_votes_ratio as ratio_c,
			  c.district_id
		  	from takattoul t join cpr c
		  	on t.district_id = c.district_id
		  ) as b
		on n.district_id = b.district_id
	  ) as d
	on circo_id = d.district_id
);

You are now ready to style your data. See the Political Parties TileMill project for more details.