Vol. 1, No. 1 | Jan – June 2017 
   
 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   96 

 

Spatial Data Analysis: Recommendations for Educational 

Infrastructure in Sindh  

Abdul Aziz Ansari, M. Abdul Rehman, Ahmad Waqas, Shafaq Siddiqui 
Department of Computer Science, Sukkur IBA, Pakistan 

aziz.mscs2013@iba-suk.edu.pk , rehman@iba-suk.edu.pk , ahmad.waqas@iba-

suk.edu.pk, shafaq.siraj@iba-suk.edu.pk   

Abstract 

 Analysing the Education infrastructure has become a crucial activity in imparting 

quality teaching and resources to students. Facilitations required in improving current education 

status and future schools is an important analytical component. This is best achieved through a 

Geographical Information System (GIS) analysis of the spatial distribution of schools. In this 

work, we will execute GIS Analytics on the rural and urban school distributions in Sindh, 

Pakistan. Using a reliable dataset collected from an international survey team, GIS analysis is 

done with respect to: 1) school locations, 2) school facilities (water, sanitation, class rooms etc.) 

and 3) student’s results. We will carry out analysis at district level by presenting several spatial 

results. Correlational analysis of highly influential factors, which may impact the educational 

performance will generate recommendations for planning and development in weak areas which 

will provide useful insights regarding effective utilization of resources and new locations to build 

future schools. The time series analysis will predict the future results which may be witnessed 

through keen observations and data collections.  

Keywords: Spatial analytics, Data Analytics, Education, GIS. 

1. Introduction 
 Education is highly significant 

element for a developing country like 

Pakistan. Keeping this fact in perspective, 

Government of Pakistan has allocated 

sufficient amount of budget for improving 

education to Grass-root level. Standardized 

Achievement Test (SAT) is a reform 

initiative, a very timely and needful strategy to 

explore the dynamics of student learning in 

Sindh province. ‘World Bank’ also recognizes 

the effectiveness of SAT. Project SAT focuses 

on attitudinal changes in teachers and students 

effective learning influenced by the 

environment and infrastructure provided on 

regional basis. This study presents the analysis 

of data collected from SAT project. According 

to several reports including SAT-I, II and III, 

the quality of education throughout the Sindh 

province is alarming. According to SAT-II 

report, a test was conducted of the students of 

class V and VIII in three subjects, i.e. Science, 

Language and Math. SAT-II results show that 

the overall average score in all subjects is 

below 30% in all regions of Sindh province, 

which definitely a crucial situation. There may 

be several reasons of such failure, like 

teachers, physical infrastructure, language 

problems or the locations of schools. 

 ASER National Report 2015 

indicates a critical education status in Sindh 

mailto:aziz.mscs2013@iba-suk.edu.pk
mailto:rehman@iba-suk.edu.pk
mailto:ahmad.waqas@iba-suk.edu.pk
mailto:ahmad.waqas@iba-suk.edu.pk
mailto:shafaq.siraj@iba-suk.edu.pk


A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   97 

 

rural areas. The report indicates that less than 

40% students capable of reading, writing 

stories in Sindhi, Urdu or English and doing 

basic mathematical operations [1]. ASER 

2014 Sindh rural report shows alarming 

situations in various aspects of its education. 

Table 1.1 shows statistics gathered from 

ASER 2014 Sindh Rural Report [2] 

 

Table 1: Learning Levels 

LEARNING LEVELS (CLASS 5) 

English 
24% can read sentences in 

English 

Urdu/Sindhi 
41% can read story in 
Urdu/Sindh 

Arithmetic 
31% can do 2-Digit division in 

arithmetic 

FACILITIES AVAILABLE FOR GOVERNMENT 

PRIMARY SCHOOLS 

Funds: 26% 

Useable 

Water: 

59% 

Boundary 

Wall: 

64% 

Useable 
Toilets:48% 

 

Reform Support Unit (RSU) also shows the 

statistics about the condition of education in 

sindh. Table 2 shows statistics of SAT 2014-

15[3]. 

 

Table 2: Content Strand Based Scores 

ClassV 

Subject  
Content 

Strand 

Conte

nt 

Stran

d 

Avera

ge 

(%) 

Subje

ct & 

Overa

ll 

Avera

ge 

(%) 

Standa

rd 

Deviati

on 

Langua

ge 

Reading 54.16 
32.81 18.6 

Writing 11.47 

Math 

Number 

& 

Operation 

18.70 

18.22 12.78 
Measurem

ent 
37.74 

Geometry 14.65 

Informati

on 

Handling 

11.56 

Science 

Life 

Science 
14.76 

15.26 11.04 

Physical 

Science 
14.49 

Earth & 

Space 

Science 

28.46 

Overall Scores (%) 
22.10 11.79 

1.1.  Geographical Information System 
 Geographical Information System 

(GIS) helps us visualize, analyse, interpret and 

understand data to reveal relationships and 

trends. According to Foorte, K.E and 

M.Lynch:“A geographic information system 

(or GIS) is a system designed to capture, store, 

manipulate, manage, and present spatial or 

geographical data” [4].In the beginning, use of 

GIS was aimed at the creation of maps only. 

The automation of paper based maps provided 

new idea of analysing data geographically 

using geometrical shapes and the 

database/linked data. This method was 

initiated by the Harvard Lab for Computer 

Graphics [5]. 

1.2.  Quantum gis 
 QGIS (Quantum GIS) is stable open 

source geographic desktop application that 

provides efficient data viewing and analysis 

capabilities. Different countries and 

organization prefer GIS based analysis of 

available data that helps them in designing 

robust policies for the future of the country. 

Various independent international works have 

been carried out in order to infer the hidden 

factors that determine the progress of the 

education system in their particular country or 

region. 

1.3.  Geostatistics  
 Statistics is the science of producing 

facts and figures based on real/sample data by 

applying some analytical methods like finding 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   98 

 

averages, correlations, regression etc. This is 

an inferential approach to make decisions. The 

merger of GIS and Statistics came with new 

dimensions of analytics. In spatial/geo 

statistical analysis objects are represented by 

basic geographical symbols like lines, points 

and polygons. GIS presents spatial 

information to have independent analysis 

based on various features that highlight hidden 

patterns within data [5].  

1.4. Time Series Analysis and 
forecasting: 

 Time series is a set of observed 

points x noted at an identified time t [6]. 

Plotted points express the growing or 

declining behaviour of data. The ordered 

series should be continuous in nature. Most of 

the time, A traditional time series is composed 

of two major components: Seasonal variation 

and Trends. Seasonal component includes 

analysis of growth or pattern in periods i.e. 

weekly, monthly, quarterly or yearly, while 

Trend component is based on linear increasing 

or decreasing trend [7]. Selection of method is 

based on the context and nature of data. Time 

Series Forecasting is a method used to predict 

the future data. Observed time series points 

x1,x2,…,xN can lead to the next possible 

trend xN+h where h (h for forecasting 

horizon) is the lead time. Most of the literature 

has divided forecasting in three general 

classes which may be used together in some 

situations.  

 Judgmental forecasts  based on 

subjective judgment or perception. 

 Univariate methods based on heuristic 

data series having some linear trends. 

 Multivariate methods based on some 

predictors  [7] 

 According to the literature review [7] 

seasonality is considered as additive if it is not 

dependent of local mean and sum of the tables 

over years’ values generally are stabilized 

to∑ it = 0. Seasonality is considered as 
multiplicative if size of the seasonal variation 

is related to the local mean and sum of the 

year’s values can be normalized by modifying 

the averagei_t=1. Regression, Moving 

average and Exponential Smoothing are some 

of the popular forecasting methods. 

1.5. Pearson Correlation Coefficient 
 According to National Council on 

Measurement in Education (NCME), 

correlation coefficient r is a numeric value that 

determines the statistical relationship or 

dependencies between two variable/attributes 

[8]. This can define the positive, negative or 

neutral effect of an attribute on other within 

the same cluster. A positive relation indicates 

that the increasing change in attribute A, 

affects the attribute B positively or 

increasingly. Negative relation indicates a 

negative or decreasing effect on attribute B 

when attribute A changes increasingly. No 

relation indicates that attribute A has no effect 

on attribute B. It measures dependencies of 

variables by value -1 <r< +1. Correlation 

coefficient r can be formulated as: 

 

r =
n ∑ xy − (∑ x)(∑ y)

√n(∑ x2) − (∑ x)
2 √n(∑ y2) − (∑ y)

2

 

 

Where n is the number of data pairs and x and 

y are variables. Measures for correlation are as 

under: 

 Correlation is said to be strong, if r≥±0.8 

 Correlation is said to be weak, if r≤±0.4 

 Correlation is said to be strong positive, if 

r is near to +1. If r=+1, then it is said to be 

a perfect positive correlation. 

 Correlation is said to be strong negative, 

if r is near to -1. If r=-1, then it is said to 

be a perfect negative correlation. 

 Correlation is said to be no-relation, if r is 

near to or is 0.  

1.6. Univariate Forecasting Methods 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   99 

 

 As our data is not seasonal or 

multivariate, we must consider available 

methods to decide which would be appropriate 

to use. Available methods for univariate 

forecasting include Seasonal Moving Average 

(SMA), General/Simple Exponential 

Smoothing (SES) and Autoregressive 

Integrated Moving Average (ARIMA). 

1.7.  Seasonal Moving Average: 
 Moving average is a method to make 

time series data smooth by taking averages. 

Average is taken with or without weight 

values. This method can be applied to forecast 

data with seasonality. Forecast is predicted by 

observing average of last‘t’ terms/seasons. 

Average moves with respect to time. Moving 

averages are calculated as F=t1+t2+…+tn. 

Seasonality may occur in data for several 

reasons like climate conditions Moving 

average is a method to make time series data 

smooth by taking averages. Average is taken 

with or without weight values. This method 

can be applied to forecast data with 

seasonality. Forecast is predicted by observing 

average of last ‘t’ terms/seasons. Average 

moves with respect to time. Moving averages 

are calculated as F= (t1+t2+…+tn)/n to 

regulate the seasonality in time series. 

Seasonality defines episodic change in data 

which may occur for several reasons like 

climatic conditions, occasions, breaks etc. It 

spreads in different time periods like weekly, 

monthly, and quarterly or as factors discussed 

above. Seasonal variations can be expressed 

as a pattern or behaviour which leads towards 

future predictions.  

1.8. General exponential smoothing 
 Exponential smoothing (also known 

as Simple or Single Exponential Smoothing) 

[9] is one of the empirical univariate 

forecasting methods [10]. Forecasting 

typically is based on smoothing averages of 

prior results. In a time series data, averages are 

supposed to be weighed by subsequent 

decaying patterns in averages. Unlike other 

univariate forecasting methods such as 

Moving Average (MA), Seasonal Moving 

Average (SMA) and Autoregressive 

Integrated Moving Average (ARIMA), 

General Exponential smoothing method is 

normally used to forecast such data where 

seasonality or trend is not present. Future 

value S_n+1is predicted by calculating weight 

average of the most recent results=aX_i+ (1-

a) S_ (n-1). a is the constraint of level of the 

series which is considered as constant at local 

element of series and gradually changes over 

time [11]. The value of alpha can be selected 

by selecting the least Root Mean Squared 

Error (RMSE) where MSE = VARIANCE 

(errors) + (AVERAGE (errors)) ^2. Staring 

value (S0) is required in this method. Several 

methods have been proposed to calculate or to 

select the starting value. According to Gardner 

[11] although there is no significant empirical 

method of taking first value for forecast, but 

taking mean of data is the popular method 

while some consider first data segment in 

series as first value for forecasting. Gardner 

states exponential smoothing better as he 

found that accuracy of this method is 

amazingly accurate and easy to implement 

[11].  

1.9. Autoregressive integrated moving 
average (ARIMA) 

 ARIMA is simplification of 

traditional Auto Regressive Moving Average 

(ARMA) model. It is used to predict future 

values from time series data. It is used mostly 

in the case where non-stationary process is 

observed which means there is a proof of 

change in mean and variance with respect to 

time. The integration of this model reduces the 

non-stationary process or non-seasonality. 

Box and Jenkins states three phases of 

ARIMA modelling: identification of time 

series properties of data, estimation of 

parameters of model and checking results of 

model with respect to hypothesis [10].   

Rest of the paper is divided into 3 sections. 

Section 2 discusses the methodology. Section 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   100 

 

3 demonstrates results and recommendations. 

Finally section 4 articulates conclusion and 

scope of future work. 

2.  Methodology 
 

 
Figure 1: Architecture Diagram 

Spatial Research framework shown in figure 1 

shows the way this research work has been 

carried out. Following sections will briefly 

discuss each of the components in brief.  

2.1. DATA PREPARATION: 
 Data preparation was a vital step of 

research. A chance of missing or unidentified 

record was present which was dealt. For 

preventing abnormal behaviours in analysis, 

data cleaning. Detected errors, altered outliers 

and filled missing data were performed. After 

removing the identified anomalies from the 

dataset, dataset was converted in a shape file. 

From shape file we extracted features which 

were appropriate to mapped against the spatial 

attributes of shape file. 

2.2. Shape File: 
 Shape file is a simple, non-

topological format for storing the geometric 

location and attribute information of 

geographic features [12]. We picked the map 

of Sindh in shape file with appropriate 

attributes to map data accordingly.  

2.3. Data Mapping:  
 Collected data is in CSV (Comma 

Separated Value) format has been mapped 

with the features of shape file (.shp). The key 

spatial attributes for analysis were district 

attribute from the collected data set while it 

was the district spatial information attribute in 

the shape file. Figure 3.2 and 3.3 shows 

subject wise results of class 5 and 8 

throughout Sindh province.  

2.4. Feature Based Analysis:  
 Focus of this work is on finding 

influencing factors for education growth by 

analysing the results and facilities that are 

being provided in different schools across 

Sindh, Pakistan. So different attributes have 

been selected which are important parameters 

for highlighting the current status of facilities. 

2.5. Data Visualization 
 For data visualization, we used SPSS 

and QGIS. Generated Scatter plots and 

histograms to visualize overall results as well 

as subject based results of class 5 and 8 

students in Languages (Sindh/Urdu/English), 

Math and Science. Visual data stated the 

factual perspective of the problem. 

2.6. Time Series Forecasting  
 Data compiled from last four year’s 

(2013-2016) SAT results. Calculated overall 

averages for 23 districts of Sindh province, 

after compiling averages, identification of 

nature of time series data was the next step. 

Overall results were placed with years which 

concluded data as univariate by nature. 

Overall scores of 23 districts were mapped on 

four years, formulated as S_ij where 

S=Overall Scores, i=1, 2, 23 (districts) and 

j=1, 2, 3, 4(years). This formulation of data 

seems like panel data. As data is univariate 

and has no seasonality and no stable trend, we 

chose general exponential smoothing method 

for forecasting formulated asS_n=aX_i+ (1-a) 

S_(n-1). Alpha is the weight chosen by 

calculating the Root Mean Square Error 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   101 

 

(RMSE) within values [13]. An optimal alpha 

value was selected from a number of 

calculation Alpha=Root (Mean Square Error), 

wherea≤0≤1. The table 3.3 shows RMSE for 

different values. 

 

Table 1.3 Rmse at Different Alpha Value 

 
Alpha RMSE 

0.1 2.61 

0.2 2.69 

0.3 2.75 

0.4 2.80 

0.5 2.85 

0.6 2.91 

0.7 2.99 

0.8 3.09 

0.9 3.21 

 

RMSE 2.61 at Alpha 0.1 is the optimal weight 

in our case for smoothing forecasting [13]. 

After data preparation, the next step was to 

select first value for forecasting as it is the 

necessity of exponential smoothing method. 

Criteria for selecting first value were 

discussed earlier. We selected average of four 

years (2013, 2014, 2015, and 2016) data as 

first value for forecasting. Found predicted 

scores for 2017 to 2021 by using simple 

exponential forecasting method 

S_n=aX_i+(1-a) S_(n-1).  

3. Results and Recommendations 
 Findings of analysis against 

geographical maps are discussed in this 

section. These findings lead us to propose 

some recommendations for authorities with 

some significant facts and figures. Most of the 

findings are based on factors derived from 

collected data.  

3.1. Performance Influencing Factors: 
 This section includes discussion 

about some factors which may cause better 

performance and some assumptions which 

study has rejected. These factors were 

identified by significant statistical analysis on 

a number of attributes like scores, gender, 

medium, infrastructure, distance/location of 

schools, Teacher’s information, student’s 

family background, surveys, etc. All attributes 

were converted into numeric form for 

correlation analysis. Hypothetically by 

considering these factors, we can improve 

student’s performance. Scores in three 

subjects Language, Math and Science 

generally influence overall performance and 

there is no need to find correlation.  Table 4 

shows correlation of different attributes with 

overall scores. 

 
Table 4: Correlation of Attributes with Overall 

Average Scores 

 
Factor 

 
Correlation 

Coefficient 

Working Male 

Teachers 

-0.041** 

Working Female 

Teachers 

0.697** 

Boundary Wall 0.590** 

Toilet 0.102** 

Drinking Water 0.508* 

Electricity 0.51** 

Computer Labs 0.571 

Qualification Academic 0.401 

Experience 0.518 

Punctuality 0.555** 

One task at a time 0.410* 

Lesson plan 0.541* 

Infrastructure 0.690 

*. Correlation is significant at the 0.05 level (2-

tailed). 

**. Correlation is significant at the 0.01 level (2-

tailed) 

 

Results in table are clearly show positive, 

negative or no relation of attributes with 

overall performance. This rejects a common 

perception that spending more budgets only 

on building physical infrastructure leads 

towards better results.  On behalf of our study 

we can say that including infrastructure, there 

are some important factors also, which may 

influence student performance. Cross-

attribute analysis is another perspective of 

analysing correlation, through which we find 

positive correlation of Math subject with 

Science subject, which may conclude that by 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   102 

 

emphasizing student’s cognitive level in math 

may increase results in science also. Table 5 

and figure 4.1 states cross-subject 

correlations. 

 
TABLE 5 CROSS-SUBJECT CORRELATIONS 

Subject Correlation coefficient 

 Language Math Science 

Language - 0.452 0.498 

Math 0.452 - 0.586 

Science 0.498 0.586 - 

 

 Study found that increasing the 

number of female teachers for lower classes, 

providing proper infrastructure with basic 

facilities (such as drinking water, electricity 

boundary wall and computer labs), appointing 

experienced teachers and implementation of 

lesson plans may cause increase in overall 

performance.  After observing a number of 

school’s data, it is found that most of the 

schools don’t have sufficient facilities and 

required infrastructure. Infrastructure plays a 

vital role in school performance, as it is a 

significant influencing factor. These facilities 

include Electricity, Toilet, Drinking water and 

some other factors. Electricity is the key 

facility for students, administration and 

teaching staff in order to run all learning 

processes smoothly within a school.  Figure 2 

depicts the electricity facility available in all 

districts of Sindh, Pakistan. The District with 

green is the one in which majority of schools 

do not have electricity facility available. On 

other hand the districts with blue Color are 

those, schools which are availing electricity 

facility. 

 Drinking Water is the most important 

facility for human being to be availed. The 

statistics in figure 3 shows that majority of 

schools in Ghotki, Dadu, Nawabshah even do 

not have facilities of drinking water. Absence 

of such basic need may result in the shortage 

of attendance which may lead to even 

catastrophic results. This facility must be 

provided in each school so that it may not 

harm the education system in these districts.   

Figure. 2: Electricity Facility 

Figure. . 3 Drinking Water Facility 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   103 

 

3.2. Time series forecasting 

 In this section, predicted scores and 

maps are discussed. Forecasting based on 

statistical analysis shows an approximate view 

of coming 6 years at district level which 

displays an insight into each district as well as 

whole province for the future. Results from 

2013 to 2021(where scores from years 2013 to 

2016 are real and from 2017 to 2021 are 

predicted) were compiled in SPSS and MS 

Excel. By using General Exponential 

Forecasting method we predicted scores for 

next five years. These predicted results were 

then mapped through QGIS. Table 6 shows 

predicted overall scores of Class 5. 

  

District 2013 2014 2015 2016 2017 2018 2019 2020 2021 

Badin 18.83 23.31 25.39 27.73 23.94 23.73 23.79 23.81 23.81 

Dadu 17.88 20.88 23.86 26.08 22.30 22.09 22.15 22.17 22.17 

Hyderabad 23.51 18.35 17.08 22.38 20.31 21.43 21.97 22.13 22.17 

Jamshoro 22.48 21.30 23.37 27.94 23.86 22.77 22.34 22.21 22.18 

Matiari 19.05 17.87 24.20 26.92 22.15 22.01 22.13 22.16 22.17 

Shaheed Benazirabad 20.80 24.30 24.42 27.52 24.35 22.96 22.39 22.22 22.18 

Tando Allah Yar 20.59 23.16 22.17 26.40 23.16 22.49 22.26 22.20 22.18 

Tando Muhammad Khan 17.18 23.98 24.93 25.30 22.96 22.37 22.22 22.19 22.18 

Thatta 15.67 19.14 22.06 25.66 20.78 21.44 21.96 22.13 22.17 

Karachi City 32.00 27.00 23.74 30.50 30.70 31.71 31.95 31.99 32.00 

Jacobabad 13.86 18.86 21.47 28.48 20.88 21.41 21.95 22.13 22.17 

Qambar Shahdadkot 14.54 18.43 19.57 24.39 19.37 20.87 21.81 22.09 22.16 

Kashmore 14.81 18.98 22.02 24.09 20.11 21.18 21.89 22.11 22.16 

Larkana 15.41 17.11 17.92 22.35 18.30 20.48 21.70 22.07 22.15 

Shikarpur 15.52 15.49 17.17 23.81 18.12 20.38 21.67 22.06 22.15 

Mirpur Khas 19.81 25.68 26.32 25.68 24.45 23.02 22.41 22.23 22.19 

Sanghar 18.03 24.00 19.88 26.26 22.13 22.05 22.14 22.17 22.17 

Tharparkar 19.05 26.79 24.17 29.33 24.96 23.18 22.45 22.24 22.19 

Umerkot 22.19 27.91 25.62 28.47 26.12 23.71 22.60 22.27 22.19 

Ghotki 14.27 17.93 22.36 25.47 20.18 21.17 21.89 22.11 22.16 

Khairpur 21.94 19.49 23.03 23.24 21.96 22.05 22.14 22.17 22.17 

Naushahro Feroze 17.05 18.52 20.06 29.14 21.37 21.65 22.02 22.14 22.17 

Sukkur 19.01 16.86 17.67 22.64 19.10 20.85 21.81 22.09 22.16 

 

Table 5: Class 5 Overall Predicted Scores 

 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   104 

 

 
Figure.  4: Class 5 Overall Predicted Scores 

 

Results shown above indicate crucial 

conditions for next 5 years, which may 

develop an assumption that the situation is 

getting worse. Class 5’s results are showing a 

slowly increasing performance in some 

districts such as Karachi, Hyderabad, Qambar 

Shahdadkot, Larkana, Kashmore, Shikarpur, 

Ghotki and Sukkur. Above mentioned 

performance, As a matter of fact, it is still an 

alarming condition which must be considered 

as emergency, and needs to be tackled with 

immediate actions, whereas class 8’s results 

indicate even worse condition. In class 8, most 

of the districts are showing a declining 

performance pattern, only a few districts like 

Karachi, Mirpur Khas and Sukkur display 

slow increasing performances, which still is a 

critical situation. Karachi is the only district 

which lies in the range of 25%-35% score, 

which if we consider as cross-match analysis 

among districts, is good, but still this is not 

overall a good result. Influencing factors 

should be considered for better performance. 

This study recommends that focus should also 

be on key cognitive development of subjects 

like Math and Science along with the 

provision of basic infrastructure throughout 

the province, which may produce better 

results. Figure 4.2.3and 4.2.4showing class 5 

and 8 performance spatially with respect to 

time. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   105 

 

 

 

 

 

 

Figure.. 5: Class 5 Scores Map 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   106 

 

4. Conclusion and 
Recommendations 

 Education is the most essential factor 

for development of a country. This factor is 

visualized graphically through geographical 

information system in all districts of Sindh, 

Pakistan. It has been observed in this research 

that even basic facilities like water, electricity 

and toilets are not available in various schools 

of Sindh, Pakistan. The task is to identify 

stimulating correlational factors that may 

cause raise in educational performance by 

analysing the datasets of schools, teachers and 

student’s results, and predicting future image 

of situation in Sindh province. The analysis 

was carried out with respect to geographical 

locations at district level. There are some 

recommendations proposed by this study with 

the statistical support which should be 

considered to produce better results in future. 

Development and upgrading of infrastructure 

with basic facilities, appointment of well 

experienced faculty equipped with latest 

technology and techniques should be 

considered. Table 7 displays key influencing 

factors found in this study. 

 
Table 7: Key Influencing Factors 

 
Attribute Correlation 

Coefficient 

Working Female 

Teachers 
0.697 

Boundary Wall 0.590 

Drinking Water 0.508 

Electricity 0.510 

Computer Labs 0.571 

Experience 0.518 

Lesson Plan 0.541 

Infrastructure 0.690 

Punctuality 0.555 

 

 Karachi, Hyderabad, Qambar 

Shahdadkot, Larkana, Kashmore, Shikarpur, 

Ghotki and Sukkur may produce better results 

in future. Future predictions are still alarming 

for critical situation of education performance 

at primary and secondary level. 

5. Future work 
 Granularity is one of the key issues in 

any geo statistical analysis. Our study’s 

granularity was at district level, which 

provides a brief overview of the case. 

Expanding the work to Taluka, UC and School 

level can shape a better insight. Socio-

economic and political issues may influence 

the scenario that needs a comprehensive 

study. Collecting spatial information of all 

schools along with geographical coordinates 

may help develop a system in order to 

combine more aspects of education. 

References 
[1] ASER, ASER National Report 2015, 

2015. 

[2] ASER, Annual Status of Education 

Report (ASER) Scale & Scope, A.S. Scope, Editor 
2014. 

[3] RSU, Standardized Achievement Test 
(SAT)-III, E.a.L.D.G.o. Sindh, Editor 2014-15. 

[4] Foote, K.E. and M. Lynch, Geographic 

information systems as an integrating technology: 

context, concepts, and definitions. Austin, 

University of Texas, 1996: p. 40-44. 

[5] Burrough, P., Environmental and 
ecological statistics. Springer, 2001. 

[6] Brockwell, P.J. and R.A. Davis, Time 

series: theory and methods2013: Springer Science 
& Business Media. 

[7] Chatfield, C., Time-series 
forecasting2000: CRC Press. 

[8] Education, N.C.o.M.i. Correlation 

Coeficient. Available from: 

http://www.ncme.org/ncme/NCME/Resource_Cen

ter/Glossary/NCME/Resource_Center/Glossary1.a

spx?hkey=4bb87415-44dc-4088-9ed9-
e8515326a061#anchorC. 

[9] Wang, S., Exponential smoothing for 

forecasting and Bayesian validation of computer 
models, 2006, Georgia Institute of Technology. 

[10] Liu, L.-M., et al., Forecasting and time 

series analysis using the SCA statistical system. 

Vol. 1. 1992: Scientific Computing Associates 
DeKalb, IL. 



A. Aziz et al. Spatial Data Analysis: Recommendations for Educational Infrastructure in Sindh       (pp. 96 - 107) 

SJCMS | P-ISSN: 2520-0755 | Vol. 1 | No. 1 | © 2017 Sukkur   107 

 

[11] Gardner, E.S., Exponential smoothing: 

The state of the art. Journal of forecasting, 1985. 
4(1): p. 1-28. 

[12] Gordon, I. and V. Monastiriotis, 

Education, location, education: a spatial analysis of 

English secondary school public examination 
results. Urban Studies, 2007. 44(7): p. 1203-1228. 

[13] Ravinder, H.V., Determining The 

Optimal Values Of Exponential Smoothing 

Constants-Does Solver Really Work? American 

Journal of Business Education (Online), 2013. 6(3): 
p. 347.