ARTIFICIAL INTELLIGENCE GROUP ASSIGNMENT PROBLEM OUTLINE FORM

L7 -ARTIFICIAL INTELLIGENCE GROUP ASSIGNMENT PROBLEM OUTLINE FORM

2 Title

- University selection Using A.I.

3. What is the problem to be solved?

- Our goal is to provide web-app for students who are planning for abroad studies. Most of the online portal and consultant portal is not showing the perfect details they are not updating regularly they just want to attract the clients for their personal growth. Helping students can be better service that we can give online.

My Project is on University Selection Using AI is mainly for those want to study abroad. Here we’ll provide a website through which they can find a University that suits the user. Nowadays lots of Ideas are available for study so people are more confused about what to choose from a bunch of options. Again many suggestions from relative or from a friend but after going there they stuck so we have set parameters in which user can fill data and our algorithm will match top 5 universities for the user and its displays in ascending order by user eligibility including the cost of living, university rank & many more things. This will add clarity to the new generation, and they can select the best path for their future endeavours, and they can make progress where their potential is valued.

4.Why are you interested in this particular problem?

Nowadays, people are rushing for abroad studies due to this hurry they end up making wrong decisions. We observed people are too afraid of taking decision when they plan to study abroad budget management, visiting lots of consultancies, Initial Fees etc. students are more depressed fighting with their goals/aims and studies they might not getting the right path due to money making era for their own benefits or growth.

5. The problem need datasets to be available? If so, which dataset is to be used?

- Yes, it should be available as a Realtime data provider API that can be integrated and used in our system. Educational Institution dataset that having all list of Universities Name (Cities, States, Countries, Fees, Course, Years of Study, Course Details) College Name that data set needs to be done.

6. Which approach is appropriate for solving that problem? Please describe exactly the steps i.e. how you are going to deal with the problem at hand?

- The best approach to solve this problem is for the best analysis of the whole system to make a robust model and web-app

Following are the steps to deal with the problem.

Step 1 – System Analysis

The Problem is that several peoples fly to abroad for further studies but know knowing the purpose. They just survive over there instead of that if they had check there eligibility that can be

more useful for them because Permanent Residence is not the key to success by growing yourself is the key to success where u can enjoy with your type of people. But no one is understanding that.

The general problem is that the colleges doesn’t explore various activities on their websites and, they don’t focus on what the students want from their educational Institution. As day by day technology and methods of teaching and studying are changing, they have to update themselves so they can provide the needs that students want.

Step 2 – Requirement of New System

The students are facing the new design of portal and contents should be made so it become helpful for the students and for the Universities. The new system will give the details of the universities. Any new update from the college, fees payments, admission procedure, rank, and all the other details through the web portal or they can use the web-app for this. These are the requirements what the students wants and according to that, the new system is being developed for them and it will be in very effective way so other colleges can use that also.so it become feasible, less time consuming for the other colleges to upgrade their portal according to the needs of the students because we will design the codes and model to consumed lesser time.

Step 3 - Feasibility Study

An important outcome of the preliminary investigation i.e. after defining the scope of the project, the determination that the system requested is feasible or not is required to done. Mainly three aspects of feasible study.

Operational Feasibility:

The operational feasibility study aspects probes whether the project is operationally feasible or not, whether the system will be used if it is developed and implemented. Not, whether the system will be used if it is developed and implemented. Our System is platform independent so we can use it in any system i.e. android, windows, Linux etc. Developed system will provide adequate throughput and all necessary information. This system is very user friendly.

Technical Feasibility:

The determination of the technical feasibility determines whether the current project is technically feasible or not keeping in mind the existing software technology and available personal.

Our project is technically feasible we will use Rest API and JAVA Hibernate Framework, Android, Python, M.L. The software and tools necessary for building the project are easily available.

Economic Feasibility:

The aspects of economic feasibility study probes whether the project is economically feasible. Whether the creation of new system has enough benefits to make the costs acceptable or not?

Step 4 – Risk Analysis

Risk identification is most important process of risk management. There are three type of risks. Technical Risk, Tools Risk, People Risk

Step5 – Data Design & Modelling

Data modelling is the process of documenting complex software system design as an easily understood diagram, using text and symbols to represent the way data needs to flow. The diagram can be used as a blueprint for the construction of new system or for re- engineering legacy model.

Traditionally, data models have been built during the analysis and design phases of the project to ensure that the requirements for new application are fully understood. A data model can be thought of as flowchart that illustrate the relationship between data.

Step 6 – Dealing with dataset

Dataset need to create or it should design in a pattern that can automatically set in database.

Step 7 – Choice of best algorithm

Choosing an algorithm wisely for our problem that can maintain the sustainability of system able to give best model.

Step 7 – Data Splitting

Depending on requirement of algorithm, we will split our data into train, test and cross validation set. The train data will train our algorithm and hyperparameter tuning will be done using cross validation data. Finally, we check the performance using test data.

Step8 – Modelling

Our model will be trained using inputs of data provided by the user and become more accurate in Predicting.

Step9 – Advance Prediction and Tough Testing

Test the performance of algorithm in Elasticity, Strength, Fast Retrieval. And check the accuracy of prediction and test it satisfy the result.

Which algorithms are planned for the application?

- K-Nearest Neighbour can help to find out the nearest value that has been entered by the user.

- Linear Regression will help us to find the best-given value and arrange it in a sequence.

- Naive Bayes is used to classifying the details. After it will depend on all at system and data if its need to modify the model to achieve accuracy of the model.

-

Which quality measures are to be used to evaluate the algorithms?

A comparison of data searching and predicting techniques is critically important in deciding which algorithm, method, or metric to use for enhanced model assessment.

1) Performance of algorithm:

• Time defines how fast our algorithm is to deliver results. Time can consume by instructions, accuracy of the instruction, processing of algorithm, effects on its runtime.

• Space define Data structures take space, Using of data structure, Short and Effective Coding.

2) Predictive Power – Our model should be precise to return the nearby expected result.

3) F1score – one indicator for precision and recall.

4) Recall/Sensitivity/True Positive Rate – It is necessary to predict a yes as yes. Otherwise it will be dicey to have a user.

5) PR curve – It is the curve between precision and recall for various threshold values.