0% found this document useful (0 votes)
58 views57 pages

DTM Full Project

Uploaded by

hlmahekagarwalla
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
58 views57 pages

DTM Full Project

Uploaded by

hlmahekagarwalla
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd

SRM INSTITUTE OF SCIENCE AND TECHNOLOGY

Ramapuram, Chennai – 600 089

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

21DCS201P – DESIGN THINKING AND METHODOLOGY


ADMISSION PREDICTOR

BATCH NUMBER :
Team Members Supervisor
KAVIYAVARSHINI C K (RA2311003020795) NAME: MR.SANTHOSH
JADDU RUPA GAYATRI
(RA2311003020819)
MAHEK AGARWALLA (RA2311003020845)
CHANDRU C (RA2311003020836)
CONTENTS
❏ Abstract
❏ Scope and motivation
❏ Introduction
❏ Objective
❏ Problem statement
❏ Software and hardware requirements
❏ Literature survey
❏ Analysis of all the 5 phases
❏ Methodology
❏ Module description
❏ Result
❏ Conclusion
❏ References
ABSTRACT
● Our web application is a machine learning-powered admission
predictor.
● The website predicts the probability of a student getting
accepted into a postgraduate degree at the Stanford University.
● The website takes inputs such as GRE, TOEFL, etc., scores of a
student and predicts the chances of admission to the
respective university.
SCOPE
The scope of building a website that predicts the chances of getting into an
Ivy League institute for graduation encompasses several components:
1. User Input:
● Academic records (GPA, standardized test scores like SAT/ACT).
● Extracurricular activities (clubs, sports, community service).
● Personal achievements (awards, recognitions).
● Letters of recommendation and personal statement evaluations.
● Demographics (geographical location, socioeconomic background).
2. Data Collection and Analysis:
● Utilize historical data from past applicants to Ivy League schools.
● Implement machine learning models that analyze correlations between applicants
profiles and acceptance rates.
● Continuously update the dataset with new admissions data to improve prediction
accuracy.

3. Prediction Algorithm:
● Develop algorithms that factor in weighted criteria based on the importance Ivy League
schools place on various elements of the application (e.g., academic performance,
extracurriculars, essays).
● Offer percentage-based predictions or risk assessments.

4. User Interface (UI):


● Create a user-friendly interface for data entry.
● Provide detailed feedback on how individual factors contribute to the overall prediction.
● Suggest actionable improvements (e.g., areas to focus on to increase admission
chances).
MOTIVATION
1.Addressing Applicant Anxiety:
● Applying to Ivy League schools can be an overwhelming process, with significant uncertainty
about the likelihood of acceptance.
● A prediction tool can help reduce this anxiety by providing some clarity and guidance.

2. Improving Accessibility:
● Many students may not have access to high-quality college counseling.
● This website can democratize access to insights typically available through expensive private
counselors, especially for underrepresented or low-income applicants.

3. Data-Driven Decision Making:


● By using data analytics, students can make more informed decisions about where to apply, how to
strengthen their applications, and what factors might be most critical in their profiles.
4. Market Demand:
● There is a growing demand for tools and resources that help students navigate the
competitive college admissions process.
● A website like this could cater to a wide audience of prospective Ivy League applicants.

5. Fostering Educational Equity:


● By making sophisticated admissions tools available online, you can contribute to leveling the
playing field for applicants from diverse backgrounds, potentially helping to increase diversity
within Ivy League institutions.

6. Encouraging Self-Improvement :
● The feedback provided by the tool can motivate students to improve their academic
and extracurricular profiles.
● This guidance can help them target specific areas for enhancement to increase their chances
of admission.
INTRODUCTION
● Our web application is a machine learning-powered admission
predictor.
● The website predicts the probability of a student getting accepted into
a postgraduate degree at the Stanford University.
● Utilizing a linear regression model, the app achieves an impressive
accuracy of 91%.
● The prediction is based on key features such as GRE scores, TOEFL
scores, CGPA, university rankings, SOP and LOR scores, and research
publication status.
● It uses the Graduate Admissions Prediction Kaggle dataset, employing
scikit-learn and linear regression for accurate predictions.
OBJECTIVE
● 400 applicants have been surveyed as potential students for Stanford University.
The university weighs certain aspects of a student's education to determine their
acceptance.

● The objective is to explore what kind of data is provided, determine the most
important factors that contribute to a student's chance of admission, and select
the most accurate model to predict the probability of admission.

● Applying for a master's degree is a very expensive and is a intensive work. With
this kernel, students will guess their capacities and they will decide whether to
apply for a master's degree or not.
PROBLEM STATEMENT
● Develop a web-based platform that uses advanced data
analytics and machine learning to predict a student's
chances of admission into Ivy League schools.

● The site will offer personalized insights based on


academic, extracurricular, and personal achievements,
helping applicants optimize their profiles and make
informed application decisions.
SOFTWARE & HARDWARE REQUIREMENTS
Software Requirements:

1. Python: Streamlit is a Python library, so Python 3.6 or later is required.

2. Streamlit: The primary framework for building the app.

3. Additional Libraries: Depending on the app’s functionality, other Python


libraries for data manipulation and machine learning might be needed.

4. Development Environment: An IDE or text editor like PyCharm, VS Code, or


Jupyter Notebook.

5. Web Browser: To access and test the Streamlit app.


Hardware Requirements:
1. Processor: A modern multi-core processor, such as Intel i5 or AMD Ryzen.
2. RAM: At least 8GB of RAM for smooth operation, especially if dealing with large datasets.
3. Storage: Sufficient storage for the operating system, development tools, and data.
4. Internet Connection: Necessary for downloading dependencies and potentially deploying the
app online.
Additional Considerations:
● Cloud Hosting: If deploying the app online, services like AWS, Heroku, or Streamlit Sharing
may be used.
● Version Control: Using version control systems like Git can be beneficial for managing code
changes and collaboration.
LITERATURE SURVEY
S.No. Title of the Paper Year Journal/Conference Name Inferences

1. Predictive Modeling in 2018 Journal of Educational Data Reviews predictive models in college
College Admissions: A Mining admissions, focusing on
Literature Review
methodologies like logistic regression
and machine learning.

2. Data-Driven Approaches 2019 Journal of Higher Education Explores how data analytics aids in
in College Admission Analytics admission decisions and highlights
Decision-Making ethical considerations in predictive
modeling.

3. Challenges and 2020 International Journal of Discusses the challenges (data


Opportunities in Building Educational Technology privacy, model interpretability) and
College Admission opportunities (transparency, access) in
Prediction Systems prediction systems.

4. User Experience and 2020 Journal of Web Design and Provides best practices for designing
Interface Design for Development user-friendly predictive websites with
Predictive Websites a focus on data presentation and
accessibility.
LITERATURE SURVEY
S.No. Title of the Paper Year Journal/Conference Name Inferences

5. Using Machine Learning 2021 IEEE Transactions on Evaluates machine


to Predict College Education learning algorithms
Admissions: An (e.g., Random
Exploratory Study Forests, SVM) for
predicting
admissions
outcomes.
6. Privacy and Ethical 2021 Journal of Ethics in Education Discusses privacy,
Considerations in consent, and bias
Predictive Analytics for issues in predictive
Education analytics for
educational
purposes.
7. Predicting Ivy League 2022 Journal of Educational Focuses on Ivy
Admissions: A Case Research and Reviews League-specific
Study admissions factors
and the holistic
ANALYSIS OF ALL THE 5 PHASES

● EXPLORE
● EMPATHISE
● EXPERIMENT
● ENGAGE
● EVOVLE
EXPERIMENT – 1
EXPLORE PHASE – STEEP ANALYSIS
ADMISSION PREDICTOR
EXPERIMENT – 2
EXPLORE PHASE – STAKEHOLDER MAPPING
ADMISSION PREDICTOR
EXPERIMENT – 3
EMPATHISE PHASE – FIELD OBSERVATION
ADMISSION PREDICTOR
EXPERIMENT – 4
EMPATHISE PHASE – DEEP USER INTERVIEW
ADMISSION PREDICTOR
EXPERIMENT – 5
EMPATHISE PHASE – PERSONA DEVELOPMENT
ADMISSION PREDICTOR
EXPERIMENT – 6
EXPERIMENT PHASE – IDEATION USING SCAMPER
ADMISSION PREDICTOR
EXPERIMENT – 7
ENGAGE PHASE – STORY TELLING
ADMISSION PREDICTOR
Admission predictor
“MANY WISH TO FORESEE THE TIME
EVERY YEAR , THOUSANDS OF HOPEFULS BEFORE IT’S TIME. BUT ARE YOU SURE TO
AWAIT TO ENTER THE GATES OF THE FORESEE WHEN THE PATH IS STILL
PRESTIGIOUS STANFORD UNIVERSITY A UNCLEAR” .HERE’S WHERE SHE GETS
VERY MUCH RENOWNED INSTITUTE WHERE HERSELF TO THINK MORE DEEP ABOUT
GREAT INNOVATORS AND THINKERS WERE HER LIFE.
BORNS

SHE NOW REALISED THAT NOW ONLY


ACADEMICS BUT ALSO THE EXTRA
ONE DAY A STUDENT NAMED SHAJINI
ACTIVITIES THAT SHAPE OUR LIFE ALONG
RIYA DISCOVERED A STRANGE NOTE WITH
THAT A GAVE A HINT ABOUT A FABLED PERSEVERANCE ,PASSION ,CONFIDENCE .
ADMISSION PREDICTOR . OUT OF WITH THAT ORAB FADED AND SHE FOUND
CURIOSITY SHAJINI TRAVELLED FOR HERSELF ON THE EDGES OF THE ENCHANTED
DAYS AND FOUND THE CRYSTAL ORAB FOREST AND THUS SHE FOUND OUT THAT
THE AUTHOR OF HER OWN DESTINY IS SHE
ON A PEDESTAL AND ECHOED SOME
HERSELF…
WORDS TO HER;
SHAJINI RIYA

HOBBIES
Demographic profile Dancing, Reading, Books, Travelling
Age-18 LIFESTYLE
Gender-Female Social interactive person, Available at
Home-Chennai social media…
Family – Nuclear Family (Unmarried)
Educational background-12th
grade(CBSE)

CHALLENGES
Tech sector is highly competitive
GOALS: and has more negative impact
To develop a software beneficial to the regarding privacy and safety
society BEHAVIOUR
MOTIVATION
Early riser, Flexible towards the
Enhance myself, attain financial stability,
care my beloved ones, earn recognition job, News reader, Clear decision
and success. making, Good in problem
solving, Prefer offline
meetings…
EXPERIMENT – 8
EVOLVE PHASE – STRATEGIC
REQUIREMENTS (ADVANCING FROM
IDEA TO REALITY)
ADMISSION PREDICTOR
Strategic Requirements
ADMISSION PREDICTOR

Key Solution Components of the Big Idea Data collection and pre- processing Performance evaluation Monitoring and updating

Capabilities Required to Deliver this Solution 1.Data engineering


Component 1. Evaluation metrics 1.Monitoring tools
2.Data privacy and security
2. Error analysis 3. Visualisation 2.Model updating
3.Data integration
3.A/B testing

Current
Organizational
Assets & capabilities to be Existing data infrastructure Existing analytics platform
Admission expertise

Leveraged

Development
Strategy to
develop this capability (if Data collection and preparation Define objectives and requirements Training and validation

needed)

Medium Low
High
Requirements and Machine learning model
cost to develop (high / med / low) Features Data privacy Integration
Infrastructure Data collection Monitoring Data collection [small data set]

Data privacy [basic


Production]

External Sources of Expertise (potential


partnership) Academic
Institutions
(Universities and research centre) Data analytics firms Edtech companies
EXPERIMENT – 9
EVOLVE PHASE – ACTIVITY SYSTEM INTEGRATION
ADMISSION PREDICTOR
Action Planning for Admission Predictor Website
Idea:
• Admission Predictor Tool for postgraduate (PG) programs, aiming to help students estimate their
chances of admission into various courses based on their academic scores and other criteria.

Objectives:
• Value: Provides an easy-to-use platform for students to make informed decisions about their
applications, increasing confidence in the admission process.
• Benefits: Streamlines the decision-making process for students, reduces uncertainty, and helps
educational institutions better anticipate applicant trends.

Responsibility:
• Team Lead: One of the students will coordinate development and team activities.
• Development Leads: Two students for front-end and back-end implementation.
• Data Analyst: One student to manage data and work on algorithm accuracy.
Implementation:
• Phase 1: Build the front-end interface and integrate data collection forms.
• Phase 2: Implement prediction algorithm and connect it to the back-end.
• Phase 3: Test the tool with sample data and user testing.

Resources:
• Tools: Web development tools (HTML/CSS/JS, Python), Database (MySQL or SQLite).
• Data: Admission records from at least 2 previous PG admission cycles.
• Hosting: A cloud-based hosting service for testing and deployment.

Completion:
• Timeline: The MVP will be completed within 8 weeks, with testing and improvements in the
following 3 weeks.
Next Steps and Refinement

Idea:
• Enhancing Algorithm Accuracy: After the initial deployment, focus on refining the prediction
algorithm by incorporating more detailed parameters like extracurricular activities and
recommendations.

Objectives:
• Value: Enhances the user experience by offering more precise predictions, giving students a
better understanding of their chances.
• Benefits: Improves user trust in the platform and expands the scope of data considered for
prediction.

Responsibility:
• Data Team: Responsible for collecting additional relevant data and fine-tuning the algorithm.

Implementation:
• Phase 1: Collect additional parameters for predictions.
• Phase 2: Update the algorithm and integrate into the website.
• Phase 3: Test accuracy with a larger sample size.

Resources:
• Tools: Data analytics software (Excel, Python libraries like Pandas).
• Expertise: Consultation with academic advisors or mentors for improving prediction metrics.

Completion:
• Timeline: 6 weeks for refining the algorithm and completing user feedback integration
EXPERIMENT – 10
EVOLVE PHASE – CHANGE MANAGEMENT TOOL
USING REVIEW
ADMISSION PREDICTOR
CHANGE MANAGEMENT PLAN
VISION (REASONS FOR STAFF ENGAGEMENT COMMUNICATE VISION FOR IMPLEMENTATION EMPOWER PEOPLE CREATE QUICK WINS
CHANGE) CHANGE PLAN FOR CHANGE
Workshops and Vision Statement: Project Timeline: Set Recognition Early User Feedback:
Training: Conduct “Empowering students with clear milestones for Programs: Implement a beta
Enhance Student workshops to data-driven insights for better data collection, model Acknowledge staff version to gather
Decision-Making: familiarize staff with educational choices. development, testing, contributions to feedback from a small
Providing accurate the new system and its and deployment encourage active group of users to
predictions helps benefits. phases. participation. refine the tool.
students make
informed choices about
their educational
paths.

Improve Efficiency: Regular Meetings: Internal Newsletters: Use Resource Allocation: Training Sessions: Success Metrics:
Streamlining the Schedule regular check- newsletters to share updates Identify necessary Provide ongoing Establish clear metrics
admission process can ins to discuss progress and success stories related to resources such as training for staff to to track the tool's
reduce time and and address concerns. the project. tools, technologies, ensure they feel success and share
resources spent on and team roles confident using the early successes with
manual evaluations. required for successful new system. stakeholders.
implementation.
VISION (REASONS STAFF ENGAGEMENT COMMUNICATE IMPLEMENTATION EMPOWER PEOPLE CREATE QUICK WINS
FOR CHANGE) VISION FOR CHANGE PLAN FOR CHANGE

Leverage Involve Key Visual Aids: Create Feedback Loops: Empower Pilot Programs:
Technology: Stakeholders: infographics to Implement Champions: Launch pilot
Utilizing machine Engage represent the feedback loops Identify and train programs in
learning to predict stakeholders in the benefits of the during testing champions within selected
admissions will development admission phases to the staff who can departments to
keep the process to ensure predictor visually. incorporate user advocate for the demonstrate the
institution their needs are suggestions into change and assist effectiveness of the
competitive and met. the final product. others. predictor quickly.
innovative.

Enhance Data- Surveys and Town Hall Risk Management: User-Friendly


Driven Decision- Feedback: Use Meetings: Host Identify potential Resources: Create Share Quick
Making: Foster a surveys to gauge meetings to risks associated user guides and Wins: Publicly
culture of analytics staff sentiment present the vision with the change FAQs to help staff recognize and
within the and gather input and engage in and develop easily adapt to the celebrate the
institution to on the change open dialogue mitigation new system. contributions and
inform future process. about the changes. strategies.
strategies. early successes of
teams involved in
the project.
EXPERIMENT – 11
EVOLVE PHASE – QUICK WINS
ADMISSION PREDICTOR
IDENTIFYING QUICK WIN FOR ADMISSION PREDICTOR
What is this Quick Win about?

• Objective: Deploy a minimal viable product (MVP) of the Admission Predictor tool to predict
postgraduate (PG) admissions based on previous data for a small set of programs.
• Scope: Pilot launch targeting around 50 prospective students across 3 PG courses to gather initial user
feedback and improve the model.

What are the success indicators? How will these be measured?

• Success Indicator 1: Tool Functionality: Does the website successfully predict admission chances for
users? (Measured through mock predictions.)
• Success Indicator 2: User Feedback: Gather qualitative feedback from users after using the tool.
(Measured through survey forms post-prediction.)
• Success Indicator 3: Predictive Accuracy: Evaluate how the prediction aligns with actual admissions data.
(Measured through simulations.)
• Success Indicator 4: Bug-free Usage: Count the number of successful, error-free predictions made by
What are the resources/staff trainings needed?

• Resource 1: Data from at least 2 past PG admission cycles.


• Resource 2: Web hosting platform for launching the MVP.
• Resource 3: Development tools (Python/JavaScript frameworks) and access to a simple database
(e.g., MySQL).
• Training: Minimal need for training; instead, focus on dividing the team into specific tasks (front-
end, back-end, data collection).

Who will lead this Quick Win implementation?

• Team Leader: One student (Team Lead) will oversee overall development and coordination.
• Tech Lead: A second student will handle back-end development, ensuring proper data processing
and model accuracy.
• UI/UX Lead: A third member will focus on the user interface and experience.
• Database Management: The fourth member will handle data input and storage for the prediction
What are the key steps needed to implement this Quick Win? What is the timeline till completion?

• Step 1: Collect past PG admissions data (2 week).


• Step 2: Develop a simple algorithm for predictions (1.5 weeks).
• Step 3: Build the MVP website with basic front-end and back-end features (4 weeks).
• Step 4: Conduct user testing with a small group of users and improve based on feedback (1 week).
• Overall Timeline: Approx. 1 month for complete deployment of the MVP.

When will be the status or progress update?

• Internal Progress Meetings: Weekly check-ins to ensure that each phase is on track.
• Development Milestones: Midway (3 weeks) and final demo (8 weeks).

When will this be completed?


How would the success be communicated?

• Internal Communication: Share results of the project with professors and mentors through a
presentation.
• External Communication: Showcase the website at the project expo or demo day.
• Group Feedback Session: Gather feedback from users and iterate for future improvements.
METHODOLGY
Define Objectives:
•Goal Setting: Assess admission chances based on student profiles.
•Target Audience: Students, counselors, institutions.
Data Collection:
•Identify Data Sources: Academic records (GPA, test scores), extracurricular activities,
recommendation letters, demographic information.
•Surveys: Create forms for users to input their details.
Data Preprocessing:
•Data Cleaning: Remove duplicates and handle missing values.
•Feature Selection: Identify key features affecting admissions.
•Encoding: Convert categorical variables to numerical format.
Exploratory Data Analysis (EDA):
•Visualizations: Explore data distributions and relationships using charts.
•Statistical Analysis: Identify correlations and trends.
Model Development:
•Choose Modeling Techniques: Use linear regression along with logistic regression or
decision trees.
•Training and Testing: Split dataset (e.g., 80-20 split).
•Model Training: Train and tune hyperparameters.
Model Evaluation:
•Performance Metrics: Use R-squared for linear regression, accuracy, precision, recall
for classification models.
•Cross-Validation: Ensure robustness.
Deployment:
•Web Development: Build the web application using Flask or Django.
•Integration: Integrate the trained model for predictions.
User Testing and Feedback:
•Beta Testing: Gather feedback on usability and accuracy.
•Iterate: Refine the model based on feedback.
Maintenance and Updates:
•Regular Updates: Continuously improve the model with new
data.
•Monitoring: Track interactions to ensure accuracy.
Documentation:
•User Manual: Provide guidance for users.
•Technical Documentation: Detail methodologies for developers.
MODULE DESCRIPTION
Streamlit: Simplifies creating and deploying interactive web apps for data science with real-time updates and seamless
integration with Python libraries.

Pandas: Facilitates efficient data manipulation and analysis using DataFrames for handling structured data.

Scikit-learn: Provides a comprehensive suite of tools for developing, training, and evaluating machine learning models.

NumPy: Offers fundamental numerical computing capabilities, including support for large, multi-dimensional arrays and
matrices.

Matplotlib/Seaborn: Enables data visualization through a range of plotting functions, helping to create informative and
attractive charts.

Joblib: Handles serialization and persistence of Python objects, particularly useful for saving and loading machine learning
models.

Requests: Manages HTTP requests for interacting with APIs and fetching data from web resources.
RESULTS
LET’S DIVE INTO THE WEBSITE!!
THE INTRODUCTORY PAGE
THE PREDICTION PAGE
THE DATASET PAGE
Our Modification: Enhancing the
Predictor's Capabilities

User Feedback
Regularly gathering user feedback to identify areas for improvement and enhance the predictor's accuracy.

Data Updates
Continuously updating the database with new data to ensure the predictor remains relevant and accurate.

Algorithm Refinement
Constantly refining the machine learning algorithms to improve prediction accuracy and personalization.
Our Innovation:
Personalized Guidance
Profile Analysis
The website analyzes students' strengths, weaknesses, and
interests to identify areas for improvement.

Targeted Suggestions
Recommendations include tailored advice on course
selection, extracurricular involvement, and essay writing.

Optimized Strategy
Students gain a clear understanding of the steps they need to
take to strengthen their applications.
CONCLUSION
● Streamlit offers a user-friendly framework for building interactive
web applications.
● Integrating your machine learning model allows for real-time
predictions.
● The platform provides tools to create an intuitive and engaging
user interface.
● Extensive documentation and community resources are available
to assist in development.
● Developing this app showcases your skills in applying machine
learning and creating practical solutions.
● The admission predictor website serves as a valuable resource for
students navigating the complex college application process. By
leveraging data, the website provides personalized insights,
guidance, and support, ultimately empowering students to achieve
their dreams and pursue their academic aspirations.
REFERENCES
● https://docs.streamlit.io/library

● https://towardsdatascience.com/a-complete-guide-to-streamlit-for-beginners-
dc43b27fc7e7

● https://medium.com/swlh/building-a-machine-learning-web-app-with-streamlit-
330e7d7c4931

● https://www.datacamp.com/community/tutorials/deploy-machine-learning-
model-streamlit

● https://www.analyticsvidhya.com/blog/2020/09/streamlit-a-powerful-library-for-
data-scientists/

You might also like