smartAd-abtesting

• SmartAd is an advertisement company.
• They Create Ads for client and charge based on user engagement.
• They also quantify the increase in brand awareness as a result of Ads shown. This is their Brand Impact Optimizer BMO.
• The company is based on the principle of voluntary participation.
• This has been proven to increase brand awareness and memorability 10X more than static options.

BMO (Brand Impact Optimizer)

you should read more on here

1. Impression
2. Reach
3. Website Traffic
4. Brand Lift
5. Avg. time on page
6. Bounce Rate SmartAd provides BMO based a lightweight questionnaire served with every campaign to determine effects on upper-funnel-metrics like memorability and brand sentiment.

The goal here is to design a reliable hypothesis testing algorithm for the BIO service, and determine whether a recent Ad campaign resulted in a significant lift in brand awareness.

---

First we will experiment with classical frequnetist techniques, and then we move on to Machine Learning based approaches.

MLOps Design

Steps to work with this repo

1. clone the repo
2. create a new environment and install the requirements.txt
3. run dvc pull to get the dataset and model files
4. use the Notebooks/logistic_regression.ipynb notebook as an example for accessing data and training a model
5. use good names for the mlflow experiment name and run names. Experiment name should be prefixed with our names.(we can discuss this)

Contributors

Made with contrib.rocks.