akhiljoe143/twitter-sentiment-analysis

---
license: apache-2.0
task_categories:
- text-classification
- token-classification
language:
- en
tags:
- twitter
- sentiment
- social
- multi-class
pretty_name: twitter-sentiment-analysis
size_categories:
- 10M<n<100M
---

# 🐦 Twitter Sentiment Analysis (bdstar/twitter-sentiment-analysis)

## 🧠 Overview
A **refined and merged version of Twitter text sentiment datasets**, providing a clean and well-balanced dataset for **sentiment classification** across three sentiment categories:
**`positive`**, **`negative`**, and **`neutral`**.

This dataset is split into three parts — **train**, **test**, and **validation** — each sourced from highly reputable open datasets.
It is designed for training, evaluating, and benchmarking **NLP models** for **Twitter Sentiment Analysis** and other **social media text classification** tasks.

---

## 🗂️ Dataset Splits

| Split | Source Dataset | Rows | File Size | Link |
|-------|----------------|------|------------|------|
| **Train** | Twitter Sentiment Dataset (3M labeled rows) | 3,142,209 | 361 MB | [Kaggle Dataset](https://www.kaggle.com/datasets/prkhrawsthi/twitter-sentiment-dataset-3-million-labelled-rows) |
| **Test** | Sentiment140 Dataset | 1,600,001 | 198 MB | [Kaggle Dataset](https://www.kaggle.com/datasets/kazanova/sentiment140) |
| **Validation** | MTEB Tweet Sentiment Extraction | 31,015 | 3.45 MB | [Hugging Face Dataset](https://huggingface.co/datasets/mteb/tweet_sentiment_extraction) |

---

## 🧩 Column Descriptions

| Column | Type | Description |
|---------|------|-------------|
| **ID** | Integer | Auto-incremental unique ID for each row |
| **text** | String | Tweet text content |
| **label** | String | Sentiment category — one of `positive`, `negative`, or `neutral` |

---

## 📊 Dataset Summary

| Property | Value |
|-----------|-------|
| **Total Rows** | 4,773,225 |
| **Columns** | 3 |
| **File Formats** | JSON / Parquet / Pandas / Polars / Croissant |
| **License** | MIT |
| **Author** | Md Abdullah Al Mamun |
| **Year** | 2025 |
| **Source** | Refined version of Twitter Sentiment Dataset |

---

## 📈 Detailed Statistics

### 🏋️‍♂️ Train Set
**Source:** [Twitter Sentiment Dataset (3M labeled rows)](https://www.kaggle.com/datasets/prkhrawsthi/twitter-sentiment-dataset-3-million-labelled-rows)
**File Size:** 361 MB
**Rows:** 3,142,209

| Label | Count | Percentage |
|--------|--------|-------------|
| Positive | 1,571,104 | 50.0% |
| Negative | 1,571,105 | 50.0% |

---

### 🧪 Test Set
**Source:** [Sentiment140](https://www.kaggle.com/datasets/kazanova/sentiment140)
**File Size:** 198 MB
**Rows:** 1,600,001

| Label | Count | Percentage |
|--------|--------|-------------|
| Positive | 800,000 | 50.0% |
| Negative | 800,001 | 50.0% |

---

### 🧭 Validation Set
**Source:** [MTEB – Tweet Sentiment Extraction](https://huggingface.co/datasets/mteb/tweet_sentiment_extraction)
**File Size:** 3.45 MB
**Rows:** 31,015

| Label | Count | Percentage |
|--------|--------|-------------|
| Neutral | 12,561 | 40.5% |
| Positive | 9,676 | 31.2% |
| Negative | 8,778 | 28.3% |

---

## 💡 Usage Example (Python)

```python
from datasets import load_dataset

# Load dataset from Hugging Face
dataset = load_dataset("bdstar/twitter-sentiment-analysis")

# Access splits
train = dataset["train"]
test = dataset["test"]
validation = dataset["validation"]

# Display sample
print(train[0])
```

---

## 🏷️ Citation
If you use this dataset in your research or application, please cite as:

```bibtex
@dataset{bdstar2025twitter,
title = {Twitter Sentiment Analysis (Refined Dataset)},
author = {Md Abdullah Al Mamun},
year = {2025},
howpublished = {Hugging Face},
url = {https://huggingface.co/datasets/bdstar/twitter-sentiment-analysis}
}
```

---

## 📬 Contact

For questions, improvements, or collaboration:
**Author:** Md Abdullah Al Mamun
📧 **Email:** [e-mail]
🌐 **Website:** [TechNTuts](https://techntuts.com/)