top of page
Search

# AI/ML Case Study Project Help | Automation in Connected Campus using AI/ML | Sample

Realcode4you experts team provide project help all AI/ML related case study topics. We are group of 15+ machine learning and data science experts that have 5+ year of experience in all AI/ML related projects. We are delivered more than 2k+ projects with 90 percent accuracy.

Below the one Sample Case Study Research Paper

• Automation in Connected Campus using AI/ML

• AI/ML for Auto-evaluate students during admission

## 1) Abstract

This project aims to create and compare different machine learning models to predict student admission to graduate school. To start, I used a tutorial from UCLA. This tutorial applies logistic regression with R, but I did it with Python. After predicting student enrollment using logistic regression, I decided to correct the unequal class distribution of the original data set.

This imbalance means it wasn't covered in the tutorial, so this was an excellent opportunity for me to correct the imbalanced dataset. After the resampling, I use logistic regression again with the new data set. To go further in the project, I not only used logistic regression but also implemented three other algorithms (decision tree, SVM-SVC, and random forest), as well as performance measurement and K-factor for both models. To summarize, I created a graph to compare the accuracy scores of different ML (Accuracy Comparison Graph) algorithms.

The primary objective is to discuss predicting college admissions based on multiple factors and using logistic regression. Many potential students apply for master's programs, and admission decisions depend on the criteria of the college or degree program. The independent variables of this study will be statistically measured to predict admission to higher education. Data mining and analysis, if successful, will allow predictive models to better prioritize the candidate selection process for a master's program, thus allowing admission of suitable candidates.

## 2) Problem Statement

A researcher is interested in how variables, such as GRE (Graduate Profile test scores), GPA (grade point average), and the reputation of the university institution, affect college admission. Graduate school. The response variable, admit/disclaimer, is binary. This dataset has a binary response variable (outcome, dependent) called admission, which is equal to 1 if the individual is admitted to graduate school and zero otherwise. There are three predictors: gre, GPA, and rank. We will treat the variables gre and GPA as continuous. The level of the variable takes values ​​from 1 to 4. Organizations ranked 1 have the highest reputation, while organizations ranked 4 have the lowest reputation

## 3) Introduction

Admission of students to a master's program includes various criteria/scores considered before admitting students to the program of study. The process is elaborate and requires a lot of thought and analysis by the selection committee before selecting suitable candidates for the master's program. This analysis aims to demonstrate that the score best contributes to the student's admission to the master's program. What factors contribute to successful access to a master's program? The analysis seems simple, but care should be taken to consider scores such as GRE, TOEFL, college rankings, SOP, LOR, and CGPA, and any outliers should not affect the decision-making process.

Students have been studying hard for a long time, and the results show they get better grades. However, the highest-ranked students may not be the best candidates for a master's program. Some form of assessment is needed to ensure that students with poor results do not have a chance to enter the program. A system that doesn't consider academic results and only considers grades will limit the ability to enroll students with other skills. Another limitation of this system is that it can also exclude people who are weak in the language of instruction or have problems related to test anxiety, among others. In such situations, the admissions committee may be asked to discuss further and consider other factors, such as the level of academic achievement in the bachelor's program. Students with weak academic ability and low test scores have been admitted to a bachelor's program, so the enrollment of such students also has many advantages. The outcome of a master's program can help students learn, grow, and progress in their academic lives. However, such a student is not required to enter a master's program.

Evaluating a student by assessing entrance exam scores is a traditional approach to considering applicants for admission. The candidate with the highest test score will be admitted. However, a student's admission decision can also be influenced by other factors, such as personality, attitude, extracurricular activities, leadership, job performance, and others. Therefore, the decision-making process must be obvious and transparent to ensure that those who score well on the test are the most suitable and qualified for the master's program.

Students who do well in school and do well in college-level subjects don't necessarily score higher on tests, as previous research has noted. These students are more likely to be admitted to a master's program even if they score lower on the entrance tests than their lower-achieving peers. To get the most out of an entrance exam, students need to follow the program they will be applying for and think about the type of score required, the skills they need to master, and the areas they need to focus on. in his research. Students can score higher on

## 4) Data Collection

The topic analysis will require data collection and generation from the UCLA graduate dataset. The existing dataset will be used to analyze and predict the factors affecting the admissions process. This dataset was created to predict graduate enrollment from an Indian point of view. Many prospective students apply for UCLA master's degree programs, and admission decisions depend on the criteria of the college or degree program. The independent variables of this study will be statistically measured to predict admission to higher education. If successful, data mining and analysis will allow predictive models to better prioritize selecting applicants to a master's program, thereby allowing admission of suitable candidates.

## 5) Extraction and preparation of data

The UCLA dataset will be examined to find predictive variables contributing to college admissions. Data cleaning will be performed to remove extraneous duplicates and outliers. The data set includes the following variables. "Table " provides detailed information on variables, whether they are predictor or response variables.

The raw data will be extracted in .csv format from the benchmark, and the RStudio import wizard will be used to import the dataset from the .csv file and perform logistic regression on the dataset. Logistic regression models the relationship between the binary response variable and the set of predictor variables. It is used to estimate the probability of a response based on different categorical and continuous predictors. The estimated probabilities can then be used to classify an unknown response into one of two outcome levels, given a set of predictors. First, we'll look for links between your predictors, such as the number of GRE, TOFEL, SOP, LOA, and Binary Response Acceptance Chance, to see which variables need to be viewed and considered for inclusion in the model. Next, logistic regression will be used to determine which students will likely be admitted to the master's program.

## 6) Analysis

• Data Handling

Showing an overview of our data

One class is dominating the other. The more is predicting more situations where the result is False. That leads to biases in the model. This model will be biased towards rejecting. I will do Part 3 anyways, only to see the results. After this I will apply a resampling method and create a new model.

Performance Measurement

The data shows that one class dominates the other. In such a case, the model will struggle to learn the data to predict future courses. Then I would apply to resample, and then I would use logistic regression again. Then I would also use three other algorithms: Decision Trees, SVMs, and Random Forests. Finally, I will do the evaluation analysis (measurement of performance and fold K)

Logistic Regression

Regression KFold

Decision Tree

SVC

## 7) Proactive suggestions for research topics to the teaching assistant

Artificial Intelligence In Education

Artificial intelligence (AI) may be applied in several sports in Education Systems; for example, comparing school, reviewing schoolwork and exams for publications may be monotonous paintings; teachers locate that reaching takes up numerous time, which would be applied to cooperate with college students, get equipped for class, or pictures at the gifted flip of events, educators can robotize reviewing for a huge variety of various choice and fill-in-the-clean testing.

Computer-primarily based intelligence can name interest to locations in which online publications want to improve, while limitless college students are observed to offer an off-base reaction to schoolwork tasks; the framework cautions the trainer. It offers destiny college students a changed message that gives symptoms of the proper answer.

Students can get more assistance from AI guides at the same time as human mentors can provide that machines can`t; few training packages depending on artificial reasoning exist and might assist college students via critical science, composing, and distinctive subjects; AI tasks can display college students basics, however, thus far are not ideal for supporting college students analyze high-quit wondering and inventiveness. Computerized reasoning can provide approximately the fulfillment of the course, all in all, it can guide educators and college students to create publications that can be redone to their requirements, and college students get all of the critical information that few faculties are utilizing.

Smart Content

According to a recent report from Digital Learning Day and the U.S. Department of Education’s Office of Innovation and Improvement, there are over 30,000 schools in the U.S. that have implemented computer-based digital learning technologies. The number continues to grow, with more than 2.3 million students in schools using digital learning technologies. A study by the United States Chamber of Commerce found that half of the students use digital technology in class. Today’s digital learning and smart content service providers are a testament to the fact that there is a need for a more streamlined and efficient way to bring all these digital services to students.

These companies can play a large role in alleviating the burden of paying for multiple student accounts and textbook/accessorized services. With smart content and digital learning technologies becoming more mainstream, the next step is bringing digital learning into secondary education. At the same time, these new technologies can become the foundation of many innovative and effective approaches to secondary education.

Content Technologies, Inc., an artificial intelligence development company specializing in business process automation and intelligent instructional design, has created a suite of intelligent content services for secondary education and beyond. So again. Cram101, for example, uses AI to help deliver and break down textbook content into an easy-to-digest “smart” study guide that includes chapter summaries, true-false, multiple-choice practice tests, and quizzes on memory cards. JustTheFacts101 has a similar, albeit simpler, goal: highlight and generate text and chapter-specific summaries, which are then stored in a digital collection and available on Amazon.

Intelligent Tutoring

The idea of building an AI tutor is prevalent among many AI systems being developed today. This system, one that would guide the AI student, is called a Student Teaching Assistant, or S/TA. A system that could provide this instruction is known as Interpersonal Communication and Pedagogy for Human-like Interactive Teaching, or ICPHEI. Research has been going on in this field for almost two decades. The AI team at IBM has been active in this area for the last couple of years and has been working on a S/TA that they have trained to achieve teaching capabilities through their Watson Conversational AI.

This system provides a fascinating example of the capabilities of ICPHEI since it has been trained using IBM’s Cognitive Toolkit (CTK). Still, it does not offer the same capabilities as ICPHEI. When I interviewed members of the AI team at IBM last year, they were quick to point out that they have been working on their S/TA since 2011 but that they only recently (as of this writing in the early summer of 2019) began testing their S/TA as it might work in the classroom. As more testing is done, it will be exciting to see how this system can adapt in the classroom and online education environments. The S/TA, however, is an example of how an ICPHEI would work. With this in mind, let’s examine a couple of articles that go into this further.

Mastery learning, a set of principles primarily associated with the work of school psychologist Benjamin Bloom in the 1970s, supports the effectiveness of personalized tutoring and classroom instruction. A program organized around student progress, combined with timely focused feedback, immediate opportunities for corrected practice, and supplemental activities, are learning approaches. Master the platform. Developing a personal tutoring system capable of providing these elements has been a coveted goal of AI researchers since the 1970s and 1980s.

Virtual Learning Experience

While it seems clear that no one in the education industry wants virtual humans to replace educators, the idea is to create virtual guides. It supports users in a variety of educational and therapeutic settings. It is a promising area of ​​development. While not yet a reality, the ultimate goal in this field is to create human-like virtual characters that can think, act, react, and interact in natural ways, responding to and using verbal and non-verbal communication

The Institute of Creative Technologies at the University of Southern California (USC) is a pioneer in creating intelligent virtual environments and applications that leverage AI, 3D games, and computer animation to develop virtual characters. Authentic and natural social interactions. USC researchers have several projects in the space that suggest applications that will emerge over the next two decades.

## Findings

The model was built of size 87.5 to predict students' enrollment status. Logistic regression was used to predict the model.

## 8) Conclusion

The results of this test seem to indicate that it contributes a lot to the response variable "Chance of admission." The higher the GRE and TOEFL scores, the higher the chances of admission. The prediction model has an accuracy of 87.5 and can be used to predict the events of entry based on the above factors. This form will be helpful for colleges to expect access and facilitate their selection process and timing. The model demonstrated that admission to a master's program depends on GRE, TOEFL, and other scores. This model could be significantly improved by collecting more data from students at different universities with similar selection criteria for selecting candidates for a master's program.

AI can create groups of students tailored to specific tasks called the Adaptive Group Format. AI application software can grade students' essays instantly. These tests are added to the central database, and future tests can be compared with previous trials in the database. Artificial intelligence in education is a computer technology that provides personalized, adaptive, and insightful teaching. A vital element of an AIED system is a domain knowledge model that provides the system's ability to perform tasks that students believe contribute to the solution. The student model provides a representation of learners' knowledge and skill development. The pedagogical model is the component that represents the teaching capability of the system, and finally, the Interface component provides the channel through which the learners and the system communicate.

AI in education is a revolutionary change. A report published by the Center for Integrative Research in Computing and Learning Sciences indicates that the next level of AI use in education has yet to be invented. Therefore, those working on AI applications should provide insights for educators and educational policymakers. While there are some downsides to using AI in the education sector, our future is that the education system should start exposing students to the kind of technology that has already begun using some AI. The impact of AI will be felt first at the lowest levels of education and gradually increase at the higher levels. The ultimate effect of AI in education will only be decided over time. AI's primary goal is facilitating educators' work but not replacing them.