ASHOKA WOMEN’S

ENGINEERING
COLLEGE
PREDICTIVE MODELING FOR H1B VISA
APPROVAL USING MACHINE LEARNING
PROJECT GUIDE BY
KAVITHA SANTHOSHI.C 182T1A0518
C.MADDILETY SRILEKHA.G 182T1AO531
Asst Professor M.Tech(Ph.D) RAMADEVI.B.R 182T1A0508
CHANDRIKA.K 182T1A0557
 CONTENTS
 Abstract
 Introduction
 Existing system
 Disadvantages
 Proposed system
 Advantages
 Modules
 Outline of the project
 Working of the project
 Architecture
 Software and Hardware Requirements
 Conclusion and future enhancement
 References
 ABSTRACT

 In our project, our goal is to predict the outcome of H1-B visa applications that are filled by
many professional foreign nationals every year.
 Here, we framed the problem as a classification problem and applied it in order to output a
predicted case status of the application.
 The input to our algorithm is the attributes of the applicant .
 INTRODUCTION

 H1-B is a type of non-immigrant visa in the United States that allows foreign nationals to work in
occupations that require specialized knowledge and bachelor’s degree or higher in the specific
speciality.
 This visa requires the applicant to have a job offer from an employer in the US before they can file an
application to the US immigration service (USCIS).
 We believe that this prediction algorithm could be a useful resource both for the future H1-B visa
applicants and the employers who are sponsoring them.
 In order to predict the case status of the applicants we will be feeding the model with the dataset which
contains the required fields by which the machine can classify the case status as certified or denied.
 EXISITING SYSTEM

 Manual power is used in order to process that applicant will certified or denied.
 They are checking each attribute of the applicant.
 When there is a huge applicants it became complex task to a person to process each attribute for
each applicant.
 DISADVANTAGES

 It is a long process takes much time to tell a particular applicant will certified or denied.
 In some cases applications are missing and in some situations applicants are denied to get visa
even though criteria for getting visa is satisfying.
 So that people who are eligible for getting visa are also denied.
 PROPOSED SYSTEM

 Using Machine learning helps to identify whether an applicant will be certified or denied.
 In Machine learning, we are using prediction algorithm to predict the output.
 In order to predict the case status of the applicant we will be feeding the model with the dataset
which contains the required fields by which the machine can classify the case status as certified
or denied.
 ADVANTAGES

 Using this model applicants can enter information and instantly get the results without delay.
 Applicants who are satisfying the criteria in our model they are definitely certified otherwise
denied.
 MODULES

 Importing libraries
 Feeding the models with dataset
 Removing outliers
 Applying Algorithm(Random Forest algorithm)
 RANDOM FOREST ALGORITHM

 A popular machine learning algorithm that belongs to the supervised learning technique.
 Used for both classification and regression problems in ML.
 Random forest is a classifier that contains a number of decision trees on various subsets of the
given dataset and takes the average to improve the predictive accuracy of that dataset.
 The greater number of trees in the forest leads to higher accuracy and prevents the problem of
overfitting.
 OUTLINE OF THE PROJECT

 Mainly prediction means it is a statement of the expected results of the experiment based on the
hypothesis.
 In our project mainly it predicts for each and every applicant who are applied for the H1B visa
will be certified or denied.
 It will be certified when all the given attributes are satisfied.
 when the criteria didn’t satisfied it will be denied
 The Main advantage of this prediction is we will get the instant results .And also it will predict
lakhs of applications at a time.
 WORKING OF THE PROJECT

 In our model we are having three files, dataset, flask app, training.
 We are feeding the Dataset file with the applicants data. Here we are storing the data in Excel
sheet.
 Here we are using flask app for creating web applications in python.
 In training file, we are importing required libraries and models, feeding the model with dataset
and applying algorithm in order to train our model.
 After training the model we got the pkl file, we are integrating the pkl file with flask app in order
to get the output page.
 When we execute the flask app we get a localhost address as shown in the picture below:
 After copying the local host address in the chrome , which we got previously then we will get the output page
as shown in the picture:
 The report tries to show the dependency of the decision on the attributes of the application. There are different types of
attributes are there it serve as an input and the output is predicted decision.

 The attributes are : 1.Select application position

2. Enter your wage
3. Enter your application year
4. Select occupation code for the employment

 Application position :Full time employment (Yes) or Part time employment (No)

 Prevailing wage: Prevailing wage for the job being requested for temporary labour condition.

 Application year: Year when petition is filed ( In between 2011-2022)

 Occupation code for employment: Occupation will be defined as different types ,like (Education , Administrative, others)
 Here we are giving the inputs to the model to predict the output:
 Here output will be predicted as follows for the given input:
 SOFTWARE AND HARDWARE
REQUIREMENTS

Hardware:
 System :Pentium IV 2.4 GHz.
 Hard Disk :40 GB.
 Ram : 512MB.
Software:

 Operating system : Windows XP/7.

 Coding Language :Python.

 IDE :Spyder, Jupyter.

 Database : CSV file, Excel sheet

ARCHITECTURE
 Conclusion and future enhancement

 Over the past decade, the demand for H-1B visa has increased a lot and it keeps getting more every
year, so the Scope of this project is to build a system that will give a hope to each individual who are
Struggling for H1-B visa in America.

 On that basis we are using the best classification technique and predicting the status of application with good
accuracy rate.Supplemental data concerning the Standard Occupational Classification (SOC) can be gathered and
used in coordination with the dataset to obtain a more comprehensive analysis of how the H-1B Visa selection
process works. By using the wage evaluations and ranges
under SOC, the wage attribute in this data set can be correctly put in to a range of salaries .

 Which can then be used to classify the visa petitions based on occupation roles rather than location wise. In
addition, other classification algorithms other than the discriminative models can be experimented with this tested
and their performances can also be analyzed.
 REFERENCES

 “H-1B Fiscal Year (FY) 2018 Cap Season,” USCIS .[Online]. Available
https://www.uscis.gov/working -united-states/temporary - workers/h-1b-speciality-
occupationsand-fashion-models/h-1b-fiscal-year-fy-2018-cap-season.[Accessed-20-Oct-2017].
 “High-skilled visa applications hit record high”, CNNMoney. [Online]. Available:
http://money.cnn.com/2016/04/12/technology/h1b-cap-visa-fy-2017/index.html. [Accessed: 20-
Oct-2017].