Text-based Emotion Classification

📝 Emotion Classification Based on Text

📘 Problem Description

The goal is to build a classification model that predicts the primary emotion associated with a text.

Each example is represented by a short text, and the label (label) corresponds to the emotion (e.g., joy, anger, sadness, etc.).

This problem belongs to the category of multi-class classification.

---

🔹 Features

• text – the textual content of the message or article

Target label:

• label – the emotion associated with the text (string)

---

📘 Input File Structure

train.csv

Contains the following columns:

• SampleID
• text
• label

Example:

SampleID	text	label
1	"I am so happy today!"	joy
2	"I feel really angry about this situation."	anger
3	"Feeling a bit sad after watching that movie."	sadness

test.csv

Contains the same columns without label, but includes SampleID.

Example:

SampleID	text
101	"What a wonderful surprise!"
102	"I can't believe this happened."

---

📤 Submission

The output file (submission.csv) must contain exactly two columns:

• SampleID
• label – the label predicted by the model

Example:

SampleID	label
101	joy
102	surprise

---

⚙️ Evaluation

Models will be evaluated using Macro F1-score:

where:

• = number of classes
• 
• 

The final score will be scaled to 0–100, so a higher F1 results in a higher score. To achieve the maximum score, F1 must be at least 0.9.

---

📊 Source

The dataset comes from Kaggle: Kaggle Emotion Dataset