📝 Student Performance Prediction

📘 Problem Description

The goal is to build a regression model to predict the final exam score (Exam_Score) based on students' academic, social, and personal factors.

The model receives a set of features and must estimate a continuous numeric value.

---

🔹 Features

Each instance contains multiple variables, such as:

• StudyHours
• Attendance
• ParentalInvolvement
• HealthStatus
• ... (other columns present in the dataset)

Target label:

• Exam_Score — numeric exam score

---

📘 Input File Structure

train.csv

Contains all features + target label.

Required columns:

• SampleID
• various features (categorical + numeric)
• Exam_Score

Example:

SampleID	StudyHours	Attendance	ParentalInvolvement	...	Exam_Score
1	3.5	High	Medium	...	78
2	1.2	Low	Low	...	55
3	4.0	High	High	...	92

---

test.csv

Has the same structure as train.csv, but without the Exam_Score column, since it must be predicted.

Example:

SampleID	StudyHours	Attendance	ParentalInvolvement	...
101	3.0	High	Medium	...
102	0.7	Low	Low	...

---

📤 Submission

The file submission.csv must contain exactly two columns:

• SampleID
• Exam_Score — model prediction

Example:

SampleID	Exam_Score
101	81.2
102	49.7

---

⚙️ Evaluation

Model evaluation uses two metrics:

• Partial RMSE — using 50% of the data
• Complete RMSE — using all the data

Then RMSE is converted into a 0–100 score using linear interpolation:

• Low RMSE → high score
• High RMSE → low score

The ideal model (RMSE = 0) achieves the maximum score of 100.

---

📊 Data Source

The dataset is generated based on the public Kaggle dataset: Student Performance Factors Dataset