Highlights

BRIEF SUMMARY 

Procedure

• Access the data from the s3 bucket and study the folder structure. Also understanding the performance data so that to understand how to relate it to the assets directory. We can do this by using the game_id. We can also append extracted starting and end frames by using game_id = folder_name, which is also a sub-string of the preview link.

• Then proceed to feature extraction. Extract important features which could be very useful in KPI determination from image files. These features are shared in the results section of this blog.

• Finally, after appending these extracted features to the performance data, and applying some pre-processing like dropping non-English text and normalizing, labeling, and scaling, we fed the data to the machine learning model, trained the model, and saved it in the models' folder. (Further reading)

BRIEF SUMMARY 

Methods for classical testings

• Defining our null and alternative hypotheses

• We defined our null hypothesis as: the change in the smartAd design which is newly made for the exposed group would result in no change in the conversion rate (i.e., number of people who knew the brand Lux would stay the same)

• Alternative hypothesis: the change in the smartAd design would result in either an increase or a decrease in awareness of the conversion rate (i.e., the number of people who knew the brand Lux would significantly either increase of decrease). We chose to use a p-value of 0.05. This means that the we will consider the ad campaign successful if the conversion rate of participants who viewed the ad was 5% more than that of those who did not view the ad.

Machine Learning Approach

• Carried out 3 types of classification analysis to predict whether a user responds yes to brand awareness,namely: Logistic Regression Decision Trees XGboost ,then compared the different classification models to assess the best performing one(s).

APPROACH

• Exploratory data analysis is the lifeblood of every meaningful machine learning project. It helps us unravel the nature of the data and sometimes informs how you go about modeling. A careful exploration of the data encapsulates checking all available features, checking their interactions and correlation as well as their variability with respect to the target. In this task, We explore the behavior of customers in the various stores. Our goal is to check how some measures such as promos and opening of new stores affect purchasing behavior.

• After cleaning the data, visualizing various features and interactions is necessary for clearly communicating our findings. It is a powerful tool in the data science toolbox. Communicate the findings below via the necessary plots. 

• Finally after training ML models using deep learning, a presentation is made through a stream lit dashboard.

Metrics:

• Response count/frequency: Combining the yes and no columns to a single response column and plotting its count gave us a great insight about how the experiment group responded against the control group.(Further reading)