Statistical Techniques for Student Grade Prediction Analysis Report

This report details a student's analysis of student grade prediction, focusing on the application of statistical techniques to a dataset of secondary school students. The analysis involved multiple regression, aiming to predict final grades based on various factors. The student's group utilized a dataset from the UCL Machine Learning Repository, employing methods like data cleaning, variable identification, and model creation. The report highlights both the strengths and weaknesses of the group's approach, including challenges in data splitting, statistical errors, and the importance of effective presentation skills. The student reflects on the learning experience, emphasizing the significance of teamwork, individual contributions, and the need for clear statistical communication. The report also addresses limitations in the analysis, such as inadequate data cleaning and the need for more comprehensive statistical interpretations.

In most learning institutions, students prefer group discussions, (Arja, et, al, 2020). However, group work
has remained a global challenge in learning institutions. Despite challenges, while working as a group,
this method remains one of the best while studying statistics-related courses. Having realized the
benefit of the group, I’m pleased to report that my group upholds togetherness while working on
different assignments. Every team member was assigned a specific assignment to work on before our
meetings and this made the group have a very productive meeting. Moreover, as a group, all members
could conduct an extensive literature search on the best statistical analysis to be conducted on our
datasets. Through the multiple regression analysis, we conducted, persuading listeners to believe on the
results of predicting student’s grades as true findings that can be generalized to other settings. Sad to
note, that most listeners get bored on the statistical terminologies like multiple regression analysis.
However, other statistical theories that were applicable and could have been done is the bivariate
analysis such as Pearson correlation and independent sample t test. For instance, an independent
sample t test would have been conducted to find out whether there is gender difference when it comes
to predicting the student’s grades. However, my group did equally well to predict student’s grades and
this made members to get an opportunity to present much information on ANOVA and even
interpretation of the p-values. Unfortunately, the statistical presentation had no explanation on the
coefficient values which I believe would have been ideal to indicate how a unit change in one variable
like education level would have resulted into a unit change in student’s grade. As human beings are to
error, in some circumstances, the group had some statistical errors in the presentation, (Strauch, 2017).
For instance, the use of boxplots to present statistical significance difference in parent’s education.
Personally, having trained myself to only give relevant information as per our arguments make me have
extreme confidence in a great job I did. Just to ensure that I avoided confusion, I only presented a few
statistics with interpretations that were relevant and avoided presentations on other matters that were
not necessary. To have produced the great presentations, I and my partners put great effort while
working on the presentation. Without wasting time, brainstorming on the statistical ideas begun
immediately after the assignment was given to every group member. To come up with persuasion to our
audiences on predicting student’s grades, we all assigned one another a specific task among ourselves.
Through the collaboration efforts, we managed to get adequate information to make our presentation
better. As much as we had some relevant statistics in the presentation, still believe that data cleaning
was appropriate something that was not equally considered. For instance, in training dataset, there
were 318 records compared to 77 records in test dataset. There was need to split this data in almost
same proportion to avoid biasness on the results. Through this exercise, we learned that every individual
has different strengths that can be tapped to make a good presentation. For instance, the presentation
was delegated to the group member with computer skills.
However, as much as we had a good presentation that was made in a PowerPoint, this is not a
guarantee of a successful speech. Therefore, it is true that a good presentation is accompanied by the
way one does the presentation and not necessarily the presentation content itself. There was needed to
also use hypothesis testing theories to prepare the audiences on what conclusions are to be made on
the student’s grade prediction. Therefore, to have avoided this mistake, there was a need to include
some hypothetical sentiments. Due to lack of descriptive statistics which are very key while preparing
the audiences, in one way or the other, it is likely to affect the flow and attention of the audiences who
lost interest in our presentation due to statistical jargons like multiple regression analysis without
informing participants what it is and why use it. Even though our presentation was having great content,
I strongly believed that more of the class members would have been convinced that our presentation

predicts students’ grades given that there is not much variation. For instance, a student in grade 3, index
six had 19 marks but this is predicted to be 18.
Moreover, our student grade prediction was rich with information and general statistics that can
convince an individual of great effort we put in place. I have a strong conviction that our group had a
statistical knowledge on student grade prediction and only highlight key statistical tests that is impactful
and persuasive. Generally, through this project, I have learned how to make a great presentation. For
instance, I have observed that the way one does presentation is even better than the content within the
presentation itself. Besides, to be a good speaker, one must do frequent practices and even rehearse
the presentation before making the actual presentation to other audiences as well. Through teamwork
and individual commitments made us have a good presentation and as a result, I have always learned to
put extra effort into any task assigned from group work.
Finally, as much as the group made a good presentation, several limitations on the statistical theories
were identified. For example, all the statistical analyses and interpretations were inadequate hence
leaving the audience blank on the interpretation of the statistical work. Also, data cleaning aspects
which are very key before writing a presentation were not done meaning that the statistical findings
may not be generalized to other different contexts.
References
Alston, P. (2017). The populist challenge to human rights. Journal of Human Rights Practice, 9(1), 1-15.
Arja, S. B., Ponnusamy, K., Kottathveetil, P., Ahmed, T. F. A., Fatteh, R., & Arja, S. B. (2020).
Effectiveness of Small Group Discussions for Teaching Specific Pharmacology Concepts.
Medical Science Educator, 1-6.
Strauch, B. (2017). Investigating human error: Incidents, accidents, and complex systems. CRC Press.