Analysis & Maps

After the selection of study area, the map showing halka boundaries is downloaded and is geo-referenced after which the shapefile of union councils is overlayed on it and is edited to make new shapefile of halka boundaries. The map is shown in figure 2.

Figure 2: Map showing the georeferenced basemap and new halka boundary

We have configured freely available open source plugin twitter2qgiswithQGIS. This plugin is still in the experimental state and requires the tweepy package to be installed manually by the user. The installation process of tweepy involves number of steps and requires other packages to be installed first. Initially a .bat file is to be created using the the path for multiple python variables.This file was named as pythonpath.bat and was run using command line (run as administrator). Then setuptools was downloaded and installed using the following command:

Finally after installing setuptools, tweepy package was installed using the following command:

This tool requires twitter access tokens and keys to pull geo-located tweets. To get these tokens and keys an app was created on apps.twitter.com and required information was retrieved. Twitter2qgis is slow and often leave qgis in the not responding state, the tool is under experiment to be used for downloading geo-tagged tweets data for districts of South Punjab. The data of the required halka’s available on the Election commission website is documented in an excel file. The data fields added in the excel file contain the total number of voters as per the previous elections, the voting percentage and then the party ranking on the basis of votes gained. The data collected can be seen in figure 15. To get the value for halka importance the values for voting percentage were used. All the values were imported into RStudio to generate a boxplot, outliers were identified and neglected in range calculation, the new range was then divided into 3 classes and the halka’s having least voting percentage were ranked 3, medium as 2 and highest were ranked 1.

This data is then combined with the shapefile on the basis of Halka names to generate layers. The tweets points shapefile is used as input for point for the Point Kernel Density tool and radius value of 8 is used to over whole study area. In order to spatial join the results of the raster generated with the Halka shapefile, we needed to generate the raster attribute table. The "generate raster attribute table" requires a single band raster with the integer type values. Now our raster was single but it had float values so we used raster calculator to multiply the values with 10000 and then used "int" tool to generate an integer value raster. The results can be seen in the figure below:

Figure 3: Showing the results of point kernel density on geo-located tweets

After this "raster to point" tool was used to generate a points layer which was then joined with the halka boundaries shapefile using spatial join with mean option. After spatially join the values, the range was divided into three classes and classes were named 1,2,3 with 1 showing the high popularity and 3 least popularity. The results are shown in the figure 4.

An online survey was conducted to get the second layer of current part popularity. For this, a simple form was made using google forms and was shared with people through social platforms like Facebook and the responses were used to get the layer shown in figure 5. Once the layers are ready, they were converted to raster layers using polygon to raster

Figure 4: Map showing the popularity of PTI according to online twitter tweets

Figure 5: Map showing the popularity of PTI according to online survey

Figure 6: Map showing the voting percentage (General Elections,2013) wise Halka ranking

Figure 7: Map showing the ranking of PTI according results of General elections 2013

tool. There was no need for using the reclassify tool as this was already done in the excel sheet. Reclassify tool is only used for the twitter popularity layer to assign classes 1,2,3. Then weighted overlay tool is used to assign weights and generate the areas of high campaign importance. the weight for different layers are shown in the table below:

Factor Weights

 Halka Importance               30%

 2013 Ranking                     30%

Current Survey Ranking     25%

Twitter Ranking                  15%

Table 1: Layers and weights used in weighted overlay tool

Figure 8: Final map highlighting the areas of high, moderate and low campaigns requirement for PTI

After getting the areas of strong, medium and low campaigns requirement, we have selected one halka of Muzaffargarh, Punjab due to the availability of the data and to keep our scope limited for this project. Area is shown in figure below. For the next step we have prepared layers to highlight the locations of high suitability together arrange an event for high campaign. For this the data-sets of parcels, roads, higher

Figure 9: Map showing the location of Muzaffargarh

order health facilities, and built-up area boundaries are used. The landuses of parcels are studied and all the landuses except vacant land, plots and open spaces are declared as unsuitable for such events. The agriculture area and orchards are specially preserved due to their high agriculture value. The figure below shows the query used to select attribute and add suitability value. After setting suitability the layer is converted into a raster and can be seen in figure 10.

Figure 10: Map showing parcels divided into suitable and unsuitable classes on the basis of their

current land-use

The attribute value for urban block name was used to identify parcels that come within the city limit and the parcels that are in the outskirts, they were merged to get 2 classes and the layer was then rasterized. This was particularly done to make a layer that gives consideration to the parking issue that is likely to rise from the high vehicle movement during such campaigns. The results can be seen in the figure 11.

Figure 11: Map showing the division of parcels in inner city limits and outskirts

To make the layer for distance from the health facilities, the health facilities were selected and converted into the points. Roads network was created and service area polygons were generated for 1 km and 3 km using the ring option for polygons. The process can be seen in the figure 12.

Figure 12: Map showing the parcels categorized on the basis of distance from higher order health

facilities

A layer for distance from primary roads was created by creatingmultiple ring buffer of 1km and 3km around the primary roads and then the parcels being intersected by the buffer layers were assigned the respective distance attributes. The layer was then rasterized and can be seen in the figure 13.

          Once all the layers were ready they were loaded into the weighted overlay tool and output was generated using equal weights for each layer. The final suitability map is shown in figure 14. The map shows that area within the built-up area and faraway fromprimary roads is not suitable for organizing a campaign event.

Figure 13: Map showing parcels categorized on the basis of distance from primary roads

Figure 14: Map showing parcels categorized on the basis of distance from primary roads