Staying Safe from Instagram Scammers and Spammers

Follow these tips to stay safe! (If you want to go straight to the data and regression click here).

Key Findings: Exercise caution when interacting with accounts with the following characteristics.

  • If the account has less than 3 followers
  • The account has made fewer than 2 posts
  • If the user lacks a bio or possesses an extensive one

    • Note: The term "Fake" is used to categorize accounts that could be scammers or spammers.

    Note: The tips provided are derived from our dataset.

If a user has less than 3 followers, there is a 79% chance the account is fake.

Account 1: Shows what an account that users would want to avoid. Low follower count


Although at the moment accepting a follow request or interacting with this type of account seems harmless. According to the FTC social media scams "The top platforms identified in these reports were Instagram (36%), Facebook (28%), WhatsApp (9%), and Telegram (7%)." This means 36% of all social media scams were from Instagram users.

___________________________________________________________________________________________________________

There's a 93% probability of an account being fake if it has 0 to 2 posts.

Account 2: Shows what an account that users would want to avoid or limit interactions.



Engage with accounts with at least 2 posts. There's a 93% probability of an account being fake if it has 0 to 2 posts. Conversely, Instagram accounts with 2 or more posts reduce the likelihood of the account being spam or a scam account by 15%.

___________________________________________________________________________________________________________

There's a 63% likelihood that an account is fraudulent if it lacks a profile biography.

Below are examples of accounts: one without a bio and another with an extensive bio.








In the realm of Instagram bios, users without a bio or those with an extensive bio containing 150+ characters may raise suspicions regarding potential scammer activity.

This plot shows the probability of an account being fake is 63% in the absence of a bio, and 66% when the bio exceeds 150 characters.


___________________________________________________________________________________________________________

Regression & Data

Using Random Forest a supervised machine learning algorithm

A random forest regression explained: Imagine you have a basket of various fruits (features) like apples, oranges, pineapples, and peaches each with different weights and sizes. Random Forest Regression is like asking a group of diverse fruit experts (decision trees) to collectively estimate the average weight of a piece of fruit based on its size, combining their expertise to give you a more accurate and reliable result. Our model's predictions have an accuracy of 93%.


Shiny Output

Creating an app we are able to select any account in our data and examine what factor the account to be labeled fake.

Break Down of an Instagram account with (really) high chance of being FAKE

The likelihood of this account being fake is 99.2%. A user might want to think twice before interacting with this account. Note: Log models are used of X.followers of 0 is actually 1 follower.

___________________________________________________________________________________________________________

Reading the Break Down: Starting from the top intercept of 49.8% which is our overall fake rate in our dataset. The plot shows a breakdown of each variable with the green bars and red bars. For example the number of posts or if the account is private.

  • Green Bars: increase in chances of being fake
  • Red Bars: decrease in chances of being fake

___________________________________________________________________________________________________________

Break Down of an Instagram account with a low chance of being FAKE

The likelihood of this account being fake is 0.06%. A user should always be cautious of random accounts, but with account is safe to interact with. Note: Log models are used of X.followers of 3.029 is actually 1016 followers.




Data Dictionary

Name Description
X.Followers Number of followers a user has Numerical
X.Posts Number of post a user has Numerical
Nums.length.username Ratio of numerical characters in username to its length Numerical
Profile.Pic User has profile picture or not Catagorical (Yes,No)
X.Follows Number of accounts a user Follows Numerical
Private Account is Private or not Catagorical (Yes,No)
Nums.Length.Fullname Ratio of numerical characters in full name to its length Numerical
external.URL Has external URL Catagorical (Yes,No)
NameIsUsername Username and full name are identical Catagorical (Yes,No)
Description Length Number of characters in Bio Numerical

Tips and Guides from Instagram

Instagram is aware of how often people fall victim to these scammers. Here are some additional help to avoid getting scammed.