COMP $2180$ Grade Prediction

This exam offers a hands$-$on experience with educational data, covering the complete cycle of data science tasks$$from preprocessing to predictive modeling.

You are encouraged to apply any data transformations and utilize any predictive modeling techniques you find appropriate. You have the freedom to explore different transformations and models to develop your predictive approach.

Your models will be judged based on the F$1$ score.

Tasks:

$1.$ Data Preprocessing and Cleaning: Address any missing or inconsistent data entries as needed. You are free to apply any data transformations to prepare the dataset for predictive modeling.

$2.$ Feature Engineering: Develop and select features that you believe will effectively predict grades. Use exploratory data analysis to guide your feature selection and engineering.

$3.$ Model Selection and Development: Choose and develop any predictive models you consider appropriate for this task. This could include, but is not limited to$,$ traditional statistical models, machine learning algorithms, or ensemble methods.

$4.$ Model Optimization: Tune your selected models to maximize the F$1$ score. Experiment with different algorithms, parameter settings, and feature combinations to find the best solution.

$5.$ Model Evaluation: Evaluate the performance of your models based on their F$1$ scores. The model with the highest F$1$ score will be considered the most successful.

You are required to prepare a report on your approach and findings using following structure.

Report

Structure: $($You may omit the ones in which you did not deal with$)$

$1.$ Introduction: Briefly introduce the objectives of your project and what you hoped to achieve with your predictive model.

$2.$ Data Preprocessing: Describe the initial state of the dataset and the specific preprocessing steps you took. Include reasons for handling missing data, any transformations applied, and how you prepared the dataset for modeling.

$3.$ Feature Engineering: Explain the features you created or selected for the model. Provide rationales for why these features were expected to influence the COMP $2180$ grades.

$4.$ Model Selection and Rationale: Discuss the models you considered and why you chose the model$($s$)$ for your final predictions. Include any relevant experiments or comparisons that led to your decision.

$5.$ Model Optimization: Outline the techniques used for optimizing your model, including any parameter tuning, algorithm adjustments, or cross$-$validation methods employed.

$6.$ Model Evaluation: Present the results based on the F$1$ score and other relevant metrics. Explain why these metrics were appropriate for evaluating model performance.

$7.$ Insights and Interpretations: Highlight the key findings from your model. Discuss any interesting patterns or insights that could be useful for academic advisors or students looking to improve their performance in COMP $2180.$

$8.$ Conclusions and Recommendations: Summarize the main conclusions of your project. Offer any recommendations based on your findings that could be implemented to improve student outcomes in the course.

Submission Requirements:

In addition to your detailed report, you are required to submit the following through the Webonline.

$1.$ Transformed Dataset:

o Upload the final version of your dataset after all preprocessing and transformations have been

applied. This will allow me to understand the data foundation upon which your models were

built.

$2.$ Predictive Model:

o Submit the final model file. If the model is coded in a script $($e$.$g$.,$ a Python $.$py file, a Jupyter Notebook $.$ipynb, or an R script$),$ make sure the code is well$-$commented to explain the function of each segment of the code.

Confirm that all files have been uploaded before the final submission.

Report Submission:

While you will upload your report online, it is also required that you submit a signed hard copy directly to me$.$ Please print out your final report, sign it to confirm the authenticity of your work, and submit it in person.

$1.$ Online Submission:

o Upload the digital version of your report through the Webonline.

$2.$ Hard Copy Submission:

o Print a hard copy of your report.

o Sign the printed report to verify that it is your own work.

o Hand in the signed report to me directly. You can submit this during my office hours or at a

designated location and time which will be communicated to you. $($Last day is $13/06/2024)$

Ensure that both the digital and signed hard copies of your report are submitted by the set deadline. Failure to submit either version on time may impact the assessment of your exam.

Grade Determination Based on F$1$ Score Performance:

Your grade for this exam will be determined by your model's F$1$ score, which will be assessed against threshold values established from the overall classroom performance. These thresholds will be set based on how well the class performs as a whole.