Tiktics - Influencer Recommendation
Data Analytics
Year
2023
Context
Academic Project [Group Project]
Role
Fullstack Developer
Python
Selenium
MongoDB
Flask
ReactJSProject Overview
A social media analytic system designed to recommend TikTok influencers and predict content performance using historical data analysis.
The Problem
Traditional social media monitoring tools like Brandwatch excel at real-time sentiment analysis but often fail to provide predictive foresight. Marketers face a significant gap when trying to justify influencer partnerships, as there is no standardized way to forecast the future engagement rate of a specific creator's upcoming content based on historical trends.
The Solution
I engineered a custom ETL pipeline using Python and Selenium to scrape high-fidelity data from over 1,000 top-tier influencers, storing the results in a flexible MongoDB architecture. To solve the prediction gap, I implemented a Statistics Prediction feature using Linear Regression, which analyzes historical engagement metrics to forecast the potential reach of future content. The system then surfaces these insights through a React-based dashboard, allowing users to search by keyword and sort by predicted engagement.
The Impact & Learning
Tiktics provides marketing teams and business partners with a strategic advantage, transforming raw scraped data into actionable predictive insights. Through this project, I gained deep expertise in handling large-scale web scraping, navigating anti-bot limits, and managing the complexities of data cleaning for machine learning models, ultimately enabling more effective and data-driven marketing campaign planning.
