nahiar/instagram_bot_detection

Instagram Fake Profile Detection Dataset

Dataset Summary

This dataset contains 5,000 Instagram profiles labeled as either fake or real, designed for binary classification tasks in social media fraud detection. The dataset provides comprehensive profile features that can be used to train machine learning models to automatically identify fake Instagram accounts.

Dataset Details

• Total Samples: 5,000 profiles
• Classes: Binary (0 = Real, 1 = Fake)
• Class Distribution: Perfectly balanced (2,500 fake, 2,500 real)
• Features: 11 profile characteristics + 1 target label
• Format: CSV file

Features Description

Feature	Type	Description	Range
profile pic	Binary	Whether profile has a picture (1) or not (0)	0-1
nums/length username	Float	Ratio of numbers to total characters in username	0.0-0.92
fullname words	Integer	Number of words in the full name	0-12
nums/length fullname	Float	Ratio of numbers to total characters in full name	0.0-1.0
name==username	Binary	Whether full name equals username (1) or not (0)	0-1
description length	Integer	Character count of profile description/bio	0-150
external URL	Binary	Whether profile contains external URL (1) or not (0)	0-1
private	Binary	Whether account is private (1) or public (0)	0-1
#posts	Integer	Total number of posts	0-7,389
#followers	Integer	Number of followers	0-15.3M
#follows	Integer	Number of accounts being followed	0-7,500
fake	Binary	Target variable - Fake (1) or Real (0)	0-1

Key Statistics

• Profile Pictures: 60% of profiles have profile pictures
• Username Patterns: Average 17% numeric characters in usernames
• Descriptions: Average 21 characters in bio descriptions
• Privacy: 23% of accounts are private
• Activity: Average 103 posts per profile
• Social Metrics:
  • Average followers: ~51K (highly variable)
  • Average following: ~481 accounts

Use Cases

This dataset is ideal for:

• Binary Classification: Train models to detect fake Instagram profiles
• Feature Engineering: Analyze which profile characteristics best distinguish fake accounts
• Social Media Research: Study patterns in fraudulent social media behavior
• Anomaly Detection: Develop unsupervised methods for identifying suspicious profiles
• Educational Projects: Learn machine learning classification techniques

Quick Start

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report

# Load dataset
df = pd.read_csv('Instagram_fake_profile_dataset.csv')

# Prepare features and target
X = df.drop('fake', axis=1)
y = df['fake']

# Split data
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)

# Train model
model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)

# Evaluate
y_pred = model.predict(X_test)
print(classification_report(y_test, y_pred))

Suggested Models

• Traditional ML: Random Forest, SVM, Gradient Boosting, Logistic Regression
• Deep Learning: Neural Networks for feature interaction modeling
• Ensemble Methods: Combining multiple algorithms for improved performance
• Unsupervised: Isolation Forest, One-Class SVM for anomaly detection

Data Quality

• ✅ Balanced Dataset: Equal representation prevents class bias
• ✅ Complete Data: No missing values
• ✅ Realistic Ranges: All features show realistic distributions
• ✅ Privacy Compliant: Only behavioral features, no personal information

Research Opportunities

1. Which profile features are most predictive of fake accounts?
2. How do fake profiles differ in posting behavior vs. real users?
3. Can feature interactions improve detection accuracy?
4. What behavioral patterns emerge in fraudulent accounts?

Citation

@dataset{instagram_fake_profiles_2024,
  title={Instagram Fake Profile Detection Dataset},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/nahiar/instagram-bot-detection}
}