Final Reflection
Khyle Capulong
2025-05-13
#Part 1 #1: Introduction This dataset, sourced from Kaggle, captures performance indicators of university students enrolled in multimedia and design-focused courses. It includes metrics such as creativity, technical proficiency, instructor evaluations, innovation, and stress levels. For this analysis, I selected the first 100 observations from the dataset to explore which variables most influence overall performance. The goal is to use multiple regression to identify the strongest predictors of academic success in a creative education setting. Source: https://www.kaggle.com/datasets/ziya07/student-performance-in-multimedia-courses
#2: Analysis
# Installing necessary packages
install.packages("tidyverse")## Installing package into '/cloud/lib/x86_64-pc-linux-gnu-library/4.4'
## (as 'lib' is unspecified)install.packages("car")## Installing package into '/cloud/lib/x86_64-pc-linux-gnu-library/4.4'
## (as 'lib' is unspecified)install.packages("ggfortify")## Installing package into '/cloud/lib/x86_64-pc-linux-gnu-library/4.4'
## (as 'lib' is unspecified)install.packages("GGally")## Installing package into '/cloud/lib/x86_64-pc-linux-gnu-library/4.4'
## (as 'lib' is unspecified)# Load necessary libraries
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.5.2 ✔ tibble 3.2.1
## ✔ lubridate 1.9.4 ✔ tidyr 1.3.1
## ✔ purrr 1.0.4
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorslibrary(car) # For VIF## Loading required package: carData
##
## Attaching package: 'car'
##
## The following object is masked from 'package:dplyr':
##
## recode
##
## The following object is masked from 'package:purrr':
##
## somelibrary(ggfortify) # For regression diagnostics
library(GGally) # For correlation plot and pair plots## Registered S3 method overwritten by 'GGally':
## method from
## +.gg ggplot2# Load dataset and selecting only the first 100 observations
data <- read.csv("creative_multimedia_student_data.csv") %>%
slice(1:101)colnames(data)[colnames(data) == "Attendance (%)"] <- "Attendance"
head(data)## Student_ID Creativity_Score Technical_Proficiency Packaging_Design_Rating
## 1 STU001 8 8 7
## 2 STU002 9 6 8
## 3 STU003 6 6 6
## 4 STU004 8 9 6
## 5 STU005 8 8 6
## 6 STU006 9 6 6
## Teacher_Rating Overall_Performance Student_Involvement Engagement_Level
## 1 8 7.75 8 7
## 2 7 7.50 9 8
## 3 7 6.25 7 6
## 4 8 7.75 9 7
## 5 8 7.50 9 7
## 6 8 7.25 8 8
## Project_Completion_Time..hrs. Collaboration_Score Attendance....
## 1 25 9 96
## 2 11 9 98
## 3 15 7 94
## 4 21 7 89
## 5 14 7 96
## 6 11 7 92
## Assignment_Score Self_Learning_Hours Innovation_Score Stress_Level
## 1 6 12 8 2
## 2 7 10 6 2
## 3 6 5 9 6
## 4 7 8 8 4
## 5 9 11 8 3
## 6 9 6 8 3
## Feedback
## 1 Needs improvement
## 2 Needs improvement
## 3 Needs improvement
## 4 Needs improvement
## 5 Needs improvement
## 6 Needs improvement# Select relevant columns
data_model <- data %>%
select(Overall_Performance, Creativity_Score, Technical_Proficiency,
Teacher_Rating, Assignment_Score, Student_Involvement,
Engagement_Level, Attendance...., Innovation_Score, Stress_Level)# Fit the multiple regression model
model <- lm(Overall_Performance ~ ., data = data_model)
# Show model summary
summary(model)##
## Call:
## lm(formula = Overall_Performance ~ ., data = data_model)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.42870 -0.22234 -0.03558 0.21388 0.53199
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.133867 0.748671 1.515 0.133
## Creativity_Score 0.240001 0.025259 9.502 2.81e-15 ***
## Technical_Proficiency 0.271197 0.025090 10.809 < 2e-16 ***
## Teacher_Rating 0.294363 0.024020 12.255 < 2e-16 ***
## Assignment_Score 0.017343 0.024412 0.710 0.479
## Student_Involvement -0.033416 0.026931 -1.241 0.218
## Engagement_Level 0.028203 0.024164 1.167 0.246
## Attendance.... 0.005095 0.005967 0.854 0.395
## Innovation_Score -0.034563 0.025024 -1.381 0.171
## Stress_Level -0.013667 0.011305 -1.209 0.230
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.2656 on 91 degrees of freedom
## Multiple R-squared: 0.8189, Adjusted R-squared: 0.801
## F-statistic: 45.73 on 9 and 91 DF, p-value: < 2.2e-16# Plot predicted vs actual
data_model$Predicted <- predict(model)
ggplot(data_model, aes(x = Predicted, y = Overall_Performance)) +
geom_point(color = "blue", alpha = 0.6) +
geom_smooth(method = "lm", color = "red") +
labs(title = "Predicted vs Actual Overall Performance",
x = "Predicted Performance",
y = "Actual Performance") +
theme_minimal()## `geom_smooth()` using formula = 'y ~ x'
# 3. Interpretation, Findings, and Conclusion The multiple regression
analysis identified three key predictors of student performance in
multimedia courses:
- Creativity (coefficient = 0.240, p = 2.81e-15)
- Technical Proficiency (coefficient = 0.271, p < 2e-16)
- Teacher Rating (coefficient = 0.294, p < 2e-16)
These variables showed strong, significant, and positive effects on overall performance. In contrast, factors such as assignment score, student involvement, attendance, innovation, and stress level had no statistically significant impact.
The scatterplot of Predicted vs Actual Overall Performance supports the model’s strength. Most points align closely with the red regression line, showing a clear correlation between predicted and actual values and indicating accurate performance predictions.
Overall, the findings suggest that success in creative multimedia education is most influenced by creativity, technical skills, and instructor evaluations, highlighting where educators and programs should focus their efforts.
4. References
Ziya07. (2023). Student Performance in Multimedia Courses [Data set]. Kaggle. https://www.kaggle.com/datasets/ziya07/student-performance-in-multimedia-courses
AI Assistance Disclosure: Portions of the R code and statistical interpretation used in this tutorial were created with the assistance of ChatGPT, a language model developed by OpenAI.
Citation: OpenAI. (2024). ChatGPT (May 16 version) [Large language model]. https://chat.openai.com/
Part 2a: About me
Hi, I’m a business student majoring in Marketing at CSUB, graduating this weekend. My current objective is to combine creative strategy with data-driven insights to solve real-world marketing problems, and my long-term goal is to work in a creative analytics role where I can apply both visual design and statistical tools to guide marketing decisions. I’m completing a Bachelor of Science in Business Administration with a concentration in Marketing, and I’ve taken coursework in advertising, digital marketing, research, and analytics. Professionally, I have nearly a year of experience as a marketing and design intern for The Plug Drink and three years with CSUB Campus Programming, where I helped organize events, lead promotional efforts, and design outreach materials. I’m experienced in Excel for organizing and analyzing marketing data, and I’ve used R in this course for regression analysis and data visualization. I’ve also explored Tableau for dashboards and have strong proficiency in Canva, Figma, Adobe Creative Suite, and Google Ads. At the start of the semester, I aimed to grow more confident with data tools—and now I feel better equipped to apply these skills professionally. I would like to pursue a career in digital marketing as a data-driven UX strategist or creative analyst.
Part 2b: Course Reflection
This course helped me understand that successful marketers are not defined by just creativity or analytics alone, but by the ability to integrate both. In the beginning, I leaned heavily on my background in design and creative content, but the assignments, tutorials, and coding challenges pushed me to approach marketing with a more analytical mindset. The wrap-up article mentioned how curiosity, collaboration, and communication are critical for today’s marketers. That really shaped how I viewed our work. I realized that marketing analytics is not just about working with data. It is about knowing how to explore data with purpose, share insights clearly, and stay open to new questions along the way.
One of the most meaningful learning experiences this semester was participating in the Plot-A-Thon at CSU Channel Islands. It gave me a glimpse into how other students and teams approach storytelling through data. I used what I had learned from class and applied it to a new environment. It motivated me to learn cluster analysis in R, which helped me understand how marketers can segment audiences more strategically. Outside of class, I also explored heat maps on my own to better visualize consumer behavior. These tools showed me the value of curiosity, and how much more there is to explore when it comes to data and decision-making.
Two things I learned that I know will be useful in the next two years are: (1) how to use clustering and regression techniques to guide targeting and insights, and (2) how to interpret and present those findings visually so that teams can take action. Two things I will remember from the last six months are how empowering it felt to complete a full R project from scratch and how collaboration can still work well even when everything is done remotely.
A strategy that helped me learn better was treating class concepts as starting points and taking time to apply them to new datasets. I did not want to just follow instructions. I wanted to test and fail and figure out where my gaps were. Working this way made the material more meaningful.
If I could do something differently, I would start applying advanced R techniques earlier. I came into this class already familiar with the basics of R, but I waited too long to build on that foundation. The good thing is that I eventually did, and I now feel more confident using it in marketing-specific ways. One thing I will remember to do in the future is stay open to learning tools that make me uncomfortable at first, because those are often the ones that lead to the biggest breakthroughs. One thing I still want to learn is how to build real-time dashboards that track digital campaign performance and generate automated reports that teams can rely on.
True or false: Marketers are good at creative work but not good with data. That is false. Today’s best marketers understand design, communication, analytics, and strategy. They need to be adaptable and fluent in both sides of the field. When creative storytelling and statistical analysis come together, marketing becomes more powerful and relevant.
The most important skill I gained from this course is confidence. I now trust my ability to look at data, explore it with purpose, and explain what it means in a way that matters. That mindset will shape how I approach my future career.
Part 3
Our group never met in person. All of our collaboration happened digitally. We assigned tasks during class sessions and used text messages to follow up on progress, clarify missing pieces, and coordinate our responsibilities. Most of the work was done remotely and at each person’s own pace. While this flexible setup worked in our favor at times, it also made it harder to track who was doing what. Looking back, I think having at least one structured online meeting would have helped improve coordination and kept everyone more consistently engaged throughout the process.
We did not visit a math tutor because we were able to troubleshoot and solve most problems by relying on each other. That experience in itself was valuable. It made us take ownership of the learning process and encouraged us to test different approaches until things worked. It also built my confidence in debugging and working with code even when there was no clear step-by-step guide.
For my role, I contributed to creating the survey, helping debug some of the R code during the early phase, designing charts to support our initial findings, and building the PowerPoint slides for our final presentation. Toward the end, my teammates focused on refining the code and improving the structure of the slides. I appreciated that we all brought different strengths to the table. Even though the process was sometimes disorganized, we managed to get it done by trusting each other to follow through.
One of the biggest challenges we faced was getting enough survey responses. That limited the quality of our dataset and made it harder to draw strong conclusions from our analysis. It also taught me how important the data collection phase is. In future projects, I would be more strategic about how to attract participants and increase response rates. If I were to do this again, I would choose a topic that is more directly relevant to marketers and easier to connect with real business problems. That would likely improve both engagement from respondents and the overall value of the findings.
This group project helped me develop more than just technical skills. It gave me a better understanding of how to communicate and collaborate effectively in an online setting. It also reminded me that even with limited resources or uneven workloads, the process of problem-solving and learning together can still be meaningful.