Optimize Advertisement Moderation

TIKI-TOKI

Introduction

In the fast-paced world of digital advertising, staying competitive means constantly innovating and optimizing ad placement strategies. The TIKI-TOKI Ad Ranking Optimization project emerged as a response to this need, with the goal of enhancing our ad placement algorithms to improve user engagement and revenue generation.

As a cross-functional team, we embarked on this project with a clear mission: to learn, adapt, and overcome challenges in the dynamic landscape of digital advertising. Along the way, we not only honed our technical skills but also gained valuable insights into the intricacies of the advertising industry. This project story captures our journey, highlighting what we learned, the hurdles we faced, and how we meticulously built and deployed our solution.

Join us as we delve into the world of ad ranking optimization, sharing the knowledge we acquired, the challenges we conquered, and the strategies we employed to create a more efficient and effective ad placement system.

Brief Biographies

Loo Ying Gene Hi I'm Gene, a snoppy-lookalike currently studying Data Science at the National University of Singapore. I love Klang !
Goh Jiun Yih Hello I'm Jiun Yih, currently an undergraduate at National University of Singapore, pursuing degree in Data Science and Analytics. I am always passionate about exploring various data.
Lim Shaun Lii Hi I am Shaun, currently pursuing a degree in Business Analytics at the National University of Singapore. I'm passionate about leveraging technology to build a better world.
Lewis Chong Li Wei Hi, I'm Lewis , currently Year 3 studying Data Science and Analytics at the National Univeristy of Singapore.As an outgoing and communicative individual, I am passionate about the exciting field of data science and machine learning, and would love to learn more through this project !

What We Learned

Technical Skills

Data Analysis and Manipulation: Learning and applying data analysis libraries such as Pandas and NumPy to preprocess and manipulate advertising data efficiently.We used it for cleaning, transforming, and exploring the raw ad data.
Machine Learning and Modelling : We took the time to understand numerous machine learning algorithms including logistic regression, decision trees,clustering to see what's best fit for the data that we have for ad ranking prediction. Then, we used the best concluded model to run do prediction and optimize ranking strategies
Version Control Using Git: We got familiar to using Git both in the terminal and in our respective prefered IDEs so that we can track and manage code changes in our code. This was able to encourage collaborative development and track the changes in our repository.

Domain Knowledge

Ad Review Performance and Moderator Metrics: We delve into the knowledge of key performance metrics in advertising, such as punish number , task complexity , and review accuracy. In terms of moderators, we understood the criterias for categorising moderators, such as productivity, handling time and accuracy. Using these knowledge, we evaluated and built the ranking models to ensure high-quality content moderation and compliance.

Challenges Faced

Time Constraints: We had a limited timeframe to both self learn all of the required knowledge and building the actual model.We had to make rapid decisions, prioritize tasks effectively, and streamline our development process to meet project milestones within the allotted timeframe. This challenge pushed us to be resourceful, adapt quickly, and optimize our workflow to deliver a high-quality solution that met our project objectives.