Kon School?

Build Status:

Server	Platform	Status
Azure Pipelines	Windows (VS 2017)
	macOS (10.13)
	Linux (Ubuntu 16.04)
AppVeyor	Windows (VS 2017)
Travis CI	Linux (Ubuntu 18.04)
GitHub Actions

QC:

Kon School? (Bengali: কোন স্কুল?, meaning "Which school?") is a recommendation system that gives secondary schools personalized compatibility scores based on user's preferences.

• Process
  • Criteria of schools
  • Inputs
  • Calculation
  • Outputs
• Data on Schools
• Running the web app locally
  • Compilation of source
  • Docker image
• REST API
  • Fuzzy AHP endpoint, /api/fahp
  • School data endpoint, /api/schools
• Acknowledgement
• Contributing

Process

Criteria of Schools

• Teacher-student ratio
• Male-female ratio (among students)
• Socio-economic status (of guardians of students)
• Distance of school from home
• Age of school (based on year of establishment)
• Difference between the age of the child and the average age of the students in the intended class (grade)

Inputs

• Crisp inputs: student data (age, sex, class (grade), location, guardian's profession)
• Fuzzy inputs: pairwise comparison between criteria of schools: 5 comparisons for 6 criteria

Calculation

After getting the inputs, the system

• uses these linguitic values to run Fuzzy Analytic Hierarchy Process (Fuzzy AHP) to evaluate weights of each criterion
• takes data from a dataset on secondary schools of Bangladesh and calculates comparative values for each criterion
• gives each school a score, which is a sum of weighted criteria-values.

For more details on calculation of weights from fuzzy inputs, please check the docs.

Outputs

• Weights of 6 criteria
• Compatibility scores for all eligible schools

Data on schools

The dataset on schools that is used for this system was compiled from several datasets that were originally collected from Bangladesh Open Data. It contains the data on 14,274 schools across the country. The SQLite database can be found here. You can also use API endpoints to access the data (view the REST API section below for further instructions).

The collection of data and its preprocessing is discussed in the docs.

Running the web app locally

There are two ways.

Compilation of source

If you have .NET Core SDK (2.x) installed, you can run the application directly from the source:

git clone https://github.com/maacpiash/KonSchool
dotnet run --project KonSchool/src/KonSchool.csproj

Now, navigate to either http://localhost:5000 or https://localhost:5001.

I recommend this method over the next one. You can download .NET Core SDK installer or binary from Microsoft.

Docker image

You may run from the docker image:

docker container run -it -p 80:80 maacpiash/konschool

The app can now be accessed from http://localhost:80. You can get instructions on installing Docker from their documentation.

REST API

In addition to the server-side-rendered pages, there are two RESTful API endpoints:

Fuzzy AHP endpoint, /api/fahp

This endpoint can be used to get results of 6 criteria by providing 5 integers (between -9 and 9, inclusive).

Examples:

• /api/fahp/0,0,0,0,0 would return equal weights for 6 criteria (0.166...).
• /api/fahp/1,2,3,4,5 would return 6 weights that are exponentially increasing.
• /api/fahp/-5,-4,-3,-2,-1 would return 6 weights that are exponentially decreasing.

For detailed information regarding the integer values, please check the docs.

School data endpoint, /api/schools

This endpoint can be used to access data of schools. Schools can be queried about by EIIN, divisions or districts.

Examples:

• /api/schools would return the data of all the schools.
• /api/schools/div/RAJSHAHI would return the data of all the schools inside the division of Rajshahi.
• /api/schools/dis/BOGRA would return the data of all the schools inside the district of Bogra.
• /api/schools/100005 would return the data of the school with EIIN 100005.

All these endpoints return response in standard JSON format.

Acknowledgement

I am very grateful to Dr. Rashedur M. Rahman (NSU ECE, Google Scholar), Professor at Department of Electrical and Computer Engineering, North South University; and our honorable instructor of the Fuzzy Logic course (CSE470). He mentored my group on this project and made sure our research work got published in Springer and presented at a prestigeous conference.

I would also like to thank my groupmates — N.S.M. Rezaur Rahman, for the development of the mathematical model; and Abdullah-Al Nahian Siraj, for his cooperation with finding related research paper.

Contributing

This project is under GNU Affero General Public License v3. Please read the contribution guideline in advance.