Fake Instagram Profile Detector

Fake Instagram Profile Detector is a Python and Flask machine learning web application that predicts whether an Instagram account is Real or Fake from its public profile metadata — profile picture presence, username/full-name digit ratios, bio length, follower/following ratio, privacy setting, and post count. Enter a username and the app fetches the profile and returns a verdict with a confidence score and a plain-English report explaining which signals drove it.

Features

What You Get  

• Detect by username only — enter a handle and the app fetches the account's public profile data automatically and analyzes it.
• RandomForestClassifier trained on 11 behavioral/profile features (profile pic flag, username/name digit ratios, bio length, external URL presence, private flag, post/follower/following counts).
• "Why this verdict?" report — alongside the Real/Fake call and confidence score, a checklist explains which specific signals (missing profile picture, digit-heavy username, thin bio, lopsided follower/following ratio, etc.) pushed the verdict.
• Animated confidence ring and a single search-box UI — no sidebar, no extra menus.
• Dark mode by default with a light-mode toggle (persisted per browser), glassmorphic gradient theme, Poppins + Inter typefaces.
• AJAX-based results — no page reload.
• Fully responsive — works on desktop and mobile.

Plus, included in the source:

• Synthetic dataset generator modeled on the well-known Kaggle "Instagram Fake & Real Accounts" feature set.
• Model training pipeline with evaluation, serialized via pickle.
• Project report covering system design, features, algorithm, and evaluation.

Why This Toolkit?  

Spotting fake or bot Instagram accounts by hand means checking follower ratios, bios, and post history one at a time — slow and inconsistent. This tool runs a trained classifier over the same signals instantly and explains its reasoning, so the verdict isn't a black box.

• No manual feature entry — just a username
• Confidence score, not just a binary label
• Human-readable explanation of the verdict

Perfect For  

• Python/ML learners and final-year students
• Social media trust & safety experimentation
• Anyone needing a quick account-credibility check
• Demonstrating a full ML pipeline: data generation ? training ? serving

Requirements

• Python 3 + Flask
• scikit-learn (RandomForestClassifier), pandas, numpy, joblib
• RapidAPI Instagram Scraper (third-party lookup — keeps a real Instagram account from being flagged)
• HTML/CSS/JS front end (Poppins + Inter, glassmorphic theme)

Instructions

• Python 3.9+ installed and on your PATH
• A free RapidAPI account and API key (for the username-lookup feature)