Decision-Support System Optimization For Advertising Industry

Improving performance of a system that is designed to compile best offerings for advertisers to broadcast their commercials. Tackling the problem of missing data through applying a data approximation technique.

Discuss an idea

All Technologies Used

Python
Python
DB2
DB2
Netezza
Netezza

Motivation

The objective was to create a decision-support system that could enable advertisers to effectively target specific audience segments with their commercials at the right time, ensuring the best ROI for advertising campaigns on television. The key task was to reconstruct the system's process to create continuous dependencies using known characteristic values at discrete points in time, enabling accurate and optimal recommendations for advertisers.

Main Challenges

Challenge 1
Incomplete Data for Audience Targeting

The client wanted to target specific audiences with particular characteristics for optimal advertisement placement. However, the system only contained partial data, making it difficult to calculate the most suitable broadcast plans and target the correct audience at the right time. Azati proposed using a data approximation technique to estimate missing values and fill in the gaps, allowing for continuous and accurate predictions of viewer characteristics at any point in time.

Challenge 2
Gaps in Time-Based Viewer Data

The system required continuous data on viewer characteristics at every point of time. However, the data was only uploaded periodically, which created gaps in the information and hindered the system's ability to make accurate predictions for the future and suggest the best possible plans for advertisers. Azati suggested applying linear regression with L2-regularization to approximate missing data, thus ensuring the system could predict viewer characteristics for future time periods and provide optimal plans for advertising placement.

Key Features

  • Data Approximation: Linear regression with L2-regularization was used to fill in missing data, ensuring the system could work with continuous data and generate accurate predictions.
  • Optimal Broadcast Plans: The system provided advertisers with the best possible plans, integrating audience characteristics, budget, duration, and timing to maximize the effectiveness of each campaign.
  • Targeted Audience Predictions: The system predicted viewer characteristics for specific programs at specific times, ensuring that advertisements were seen by the right audience.
  • Real-Time Recommendations: The system could generate optimal plans for advertisers in real-time, helping them make quick decisions based on the available data.

Our Approach

Data Analysis and Pattern Identification
We analyzed the available data over multiple periods to understand the patterns and gaps in the viewer characteristics.
Regression Model Implementation
We used linear regression with L2-regularization to approximate the missing data and provide more accurate insights about viewer characteristics at discrete time intervals.
Continuous Prediction and Optimization
The optimized system was able to continuously restore the relationships between characteristics, allowing it to make predictions and offer optimal broadcast plans for advertisers at any given moment in time.

Project Impact

The optimization of the decision-support system significantly improved its performance, enabling the client to offer better-targeted and more efficient advertising solutions. By using data approximation, the system became capable of providing optimal recommendations even when complete data wasn't available. This enhanced the client’s ability to deliver highly effective advertisements, thus increasing customer satisfaction and ROI for advertisers.

Ready To Get Started

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.