Hybrid-Training & Placement Management with Prediction System

09 Issue:

2022 www.irjet.net

and Technology (IRJET)

e ISSN: 2395 0056

ISSN: 2395 0072

Hybrid-Training & Placement Management with Prediction System

Ipshita Das1 , V Akshid2 , Shashi Kant Mishra3

1,2B.Tech Student CSE Department, Bhilai Institute of Technology, Raipur, India

3AssistantProfessor, CSE Department, Bhilai Institute of Technology, Raipur, India

***

Abstract Training and management of placement is a crucial part of an educational institution in which most of the work is done manually. Manual system in thecollegesrequires a lot of manpower and time. Hence this hybrid system creates student and company databases. It can be accessed throughout the college with proper login provided. The developed system can guarantee to keep the records are safe. A placement predictor is to be designed to calculate the possibility of a student being placed in a company, subject to the criterion of the company. To achieve a better prediction system, we intend to use Machine Learning and here we are using prediction models like LogisticRegression, DecisionTree Classifier, Unsupervised learning which will cause a continuous accuracy growth.

Key Words: Machine Learning, Logistic Regression, Decision Tree Classifier, Unsupervised Learning.

1. INTRODUCTION

Placementsareconsideredtobeveryimportantforeachand everycollegeasitisthemostimportantpartofstudent’slife. Thebasicsuccessofthecollegeismeasuredbythecampus placementofthestudents.Everystudenttakesadmissionto thecollegesbyseeingthepercentageofplacementsinthe college. Placement managementsystemhelpsthe training andplacementofficerstoovercomethedifficultyinkeeping records of hundreds and thousands of students and searching the eligible students for recruitment, based on variouseligibilitycriteriaofdifferentcompanies.Ithelpsin effective and efficient utilization of the hardware and the softwareresources.Italsomakestheplacementprediction basedonthestudentsrecordsandhelpsthemanagementto clustertheplacedstudentsbasedontheirstreamsusingK means clustering. The project facilitates user friendly, reliableandfastmanagementsystem.Theplacementofficer itself can carry out operations in a smooth and effective manner. The college can maintain computerized records thusreducingpaperwork,timeandmoney.

1.1 SCOPE OF THE PROJECT

Every university provides campus placement opportunity foreachandeverystudenthowevermanyofthemjustkeeps the record in just excel sheets or paper, which on time requiredcannotperformabetterstatisticalanalysis.

Hencethisprojectisveryusefulforthosesituationswhen weneglecttobearisen.

2. PROBLEM STATEMENT

With the conventional method of keeping records of the studentsandmaintaining,itbecomeshecticwhenitcomes to predict the analysis of future growth or the number of studentscanbeplacedatthatmoment.Theoldermethods caneithermanagetheMISorpredictstheplacement

3. METHODOLOGY

TrainingandPlacementmanagementsysteminauniversity isveryimportantandessentialparttobeimplementedon MIS. This provides the TPO cell various features which completelyfacilitatesforeaseofwork.

3.1 FEATURES

3.1.1 Authentication

Authentication is the process which has to be primarily installedineachandeveryapplication,website,etc.

Thisfeatureinvolvesthreetypesofauthenticity:

1.admin/TPOhead

2.DepartmentalTPO

3.Student

3.1.2 MIS Functionalities

ThefunctionalitiesofthishybridanddynamicMISdealwith CRUDoperationofJobs,Trainings,Students,Departmental TPO,Alumni,PlacementRecords,etc.

The TPO/admin can add, read, update and delete all the possiblefunctionalitiesthatatrainingandplacementsystem requires by accessing the data entered by the student and analyzingthestatisticsbasedonthedataonrecords.

Factor value:

9001:2008

ISSN:

Fig 1:DataFlowDiagram

3.1.3 Prediction System

Inthispaperweusedvariousmachinelearningtechniquesto predict a particular student placement based on data available in the data set entered and updated while registeringbythestudent.Anotherpredictionisbasedonthe total number of students getting placed in the upcoming yearsbylearningthegraphfrompreviousyears.

ThemodelswhichweusedforboththepredictionareSVM, Logistic Regression, Random Forest Classifier, XGBoost Classifier.

PrimarilythedataisgatheredfromtheT&Pdepartmentand preprocessedthedata.Afterthatthepreprocesseddatawas againprocessedforrelevantprediction.Thenthedatawas split into training and testing following the models were trainedandtested.

a LOGISTIC REGRESSION

Logisticregressionisamachinelearningmodelwhichisused forbinaryclassificationproblems,likewhetherthestudent willacquireplacementorheneedtoworkhardontheplace heislaggingbehind.Ittakesinputvariousfactorsandbound togivebinaryoutputbetween0and1.

Logistic function = 1/(1+e x)

Thisisthelogisticequationinwhichxistheinputvariable afterprocessingitwillgivebinaryoutput.

b. DECISION TREE CLASSIFIER

This model is a supervised learning model used for both classificationandregressionproblems.Itistreestructured classifierinwhichinternalnodesrepresentsthefeatureof thedataset,wherebranchesrepresentthedecisionsandleaf nodeprovidestheoutcome.

Itthejustsimilartothetreewheretheprocessstartswith therootandfinishesintheleafbasedonthedecisionstaken anddividingthetreeintosub trees.

c. RANDOM FOREST

Random forest model is very popular supervised machine learning algorithm used for classification as well as regression. This model is capable of combining various classifiers and solve a complex problem and improves the performance.

Thismodelconsistsofvariousdecisiontreeonvarioussubset ofthedatasetprovidedandtakestheaveragetogivebetter output. Instead of trusting on decision trees it takes predictionsfromeachtreewithmaximumvotesforthefinal output.

d XGBOOST CLASSIFIER

XGBoost,whichstandsforExtremeGradientBoosting,isa scalable,distributedgradientboosteddecisiontreemachine learninglibrary.Thisisusedfortheproblemslikeregression, classification,andrankingproblem.

Thisalgorithmfindspatterntotraininadatasetwithlabels and features and predict the labels on a new dataset’s features.

e SUPPORT VECTOR MACHINE

This supervised learning model is primarily used for classification problems. This algorithm uses decision boundarythatdividesn dimensionalspaceintoclassessowe caneasilyputupcomingdataintocorrectcategory.

Labeleddataisusedtotrainthemodel.Themodelisusedfor predictionandhencetheoutputispredicted.

f LINEAR REGRESSION

Thismachinelearningmodelisbasedonsupervisedlearning andusedforregressiontask.Thispredictstheoutputbased on the values of independent variables. It finds the relationshipbetweenvariableandoutput.Theoutputofthe targetedpredictioniscompletelydependentonthenumber ofindependentvariablesandrelationshipsbetweenthem.

Linear Equation = Y = a + bX

Where, Y=dependentvariable X=Explanatoryvariable

Fig -2:Logisticregression

Fig 3:Linearregression

Thisallarethemodelswhichweimplementedinthispaper andworkedontogetvariouspredictionscoresandfuture total number of students getting placed in the upcoming years.

ThishybridprojectisveryrareinwhichboththeMISalong withmachinelearningpredictionmodelisbeenimplemented forproperunderstandingofthestatisticsoftheindividualas well as organizational growth in terms of placement and understanding the barrier which is affecting the student placements.

4. RELATED WORK

Farheen Taqi Rizvi, Naushin A. Khan, Saurabh Upadhyay, SonaliSuryawanshi,ShiburajPappu [2] workedoncreating the full stack for MIS of student management system with properauthenticationandworkingoftheproject.

IreneTreesaJose,DaibinRaju,JeebuAbrahamAniyankunju, JoelJames,MereenThomasVadakkel [5] didtheirstudyon studentplacementpredictionusingvariousmachinelearning supervised models or algorithms. They derived the conclusionthatSVMmodelwasaccurateabout100accuracy score followed by Logistic Regression, which acquired accuracyscoreof97.5.

PothugantiManvitha,NeelamSwaroopa [8] madethesame study on the student placement prediction with implementation of two algorithms Random Forest with accuracy of 86 score and Decision tree with 84 accuracy score.

5. RESULT

This project is completely made in python language and Djangoasawebframeworkusedtodesignthefrontendand linking to the back end. For the database SQLite is used whichisdefaultintheDjangosetup.

Along with the Management Information System the prediction part is also done with the help of python, csv formatdataset,sklearnlibraryformachinelearningmodels.

Factor

e ISSN: 2395 0056

ISSN: 2395 0072

The datasets is real dataset of our college placement of various batches that took part into campus or ofcampus placements.

6. FUTURE ENHANCEMENT

Asthedatagoesonincreasingintothemanagementsystem thepredictionmodelswillbetrainedwithmoreprecession forgrowthinaccuracyscoreasitisprovidingnow.

IntheStudentTrainingandPlacementmanagementsystem wecanmodifyaccordingtotheneedsoftheadministration.

7. CONCLUSIONS

TrainingandPlacementmanagementwithpredictionsystem isuniquemethodofimplementingadynamicManagement informationsystemintoany technology.Thiscanwork as recording data, registrations, authentication, dynamically performing all the task that a T&P cell of a university requires. As of previous studies and paper they have implementedbothofthe work separately andnoneofthe paper implemented both the methods together which we implemented.

As of the prediction part we had the real dataset of placement in our college of which we had implemented supervisedmachinelearningmodelslikeLogisticregression, Random Forest, Support vector machine, Decision tree classifier,XGBoostclassifierinwhichthebestscoreprovided wasbyXGBoostandDecisiontreeclassifiermodel.Theleast accuracy score in our case was given by Support vector machinemodel.

8. REFERENCES

[1] Online Training and Placement Management System, Santhosh Kumar H, Mrs. Srividhya V R InternationalResearchJournalofEngineeringand Technology(IRJET),Volume:04Issue:22|2016.

[2] PLACEMENT MANAGEMENT SYSTEM 1Farheen TaqiRizvi,1NaushinA.Khan,1SaurabhUpadhyay, 2Sonali Suryawanshi, 3Shiburaj Pappu Journal of Emerging Technologies and Innovative Research (JETIR), Volume8,Issue4 |Apr 2021.

[3]

STUDENT PLACEMENT PREDICTION USING MACHINE LEARNING Shreyas Harinath 1, Aksha Prasad2,SumaHS3,SurakshaA4,TojoMathew5 InternationalResearchJournalofEngineeringand Technology(IRJET)Volume:06Issue:04|Apr2019

[4] A Review: Placement Prediction using Machine LearningProf.SalveB.S1ShivajiWaghmode2Amol Kharade3 Manoj Kharade4 Sorate Sajjan5 International Journal for Scientific Research & Development|Vol.7,Issue09,2019

[5] Placement Prediction using Various Machine Learning Models and their Efficiency Comparison IreneTreesa Jose1,DaibinRaju2,JeebuAbraham Aniyankunju3, Joel James4, Mereen Thomas Vadakkel 5 International Journal of Innovative ScienceandResearchTechnologyVolume5,Issue5, May 2020.

[6] Placement Management System for Campus RecruitmentAjeenaSunny1AneenaFelix1Angelin Saji1 Christina Sebastian1 Praseetha V.M2

International Research Journal of Engineering and Technology (IRJET)

Volume: 09 Issue: 04 | Apr 2022 www.irjet.net

International Journal of Innovative Science and Research Technology Volume 5, Issue 5, May 2020.

[7] Placement prediction system using machine learning Rathi Viram, Swati Sinha, Bhagyashree Tayde , Aakshada Shinde International Journal of CreativeResearchThoughts(IJCRT)Volume8,Issue 4April2020.

[8] Campus Placement Prediction Using Supervised MachineLearningTechniquesPothugantiManvitha NeelamSwaroopa InternationalJournalofApplied EngineeringResearchISSN0973 4562Volume14, Number9(2019).

2022, IRJET

e ISSN: 2395 0056

p ISSN: 2395 0072

Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal