Hello :) Today is Day 217!
A quick summary of today:
- talking about the KB project
- reading papers related to bank telemarketing models
Today my KB project partner and I met to catch up. We looked over what he had done during the past week. In addition, we talked about what happens after developing a model in a jupyter notebook - creating a script that takes a new row of features, preprocessing them, and then putting that through the model to get a prediction for that new case. Then we went over the below graph
Giving him an overview of what each technology does. He is going on vacation from tomorrow to Friday, so the tasks left are to write up the project ppt that is needed for the submission in Korean.
In addition, for reproducability - I noticed that the kafka-producer uses local data to send transactions. For better reproducability, I uploaded the dataset to HuggingFace and used in the code I am reading the csv from the HF url
I also read some papers related to bank term deposit telemarketing
A dynamic ensemble selection method for bank telemarketing sales prediction
The paper proposes a method called META-DES-AAP to predict the success of bank telemarketing campaigns for selling time deposits. This method stands out by not only focusing on prediction accuracy but also on maximizing average profit. It uses a multi-objective optimization algorithm to select and integrate the most suitable base classifiers dynamically for each campaign. The results show that META-DES-AAP outperforms other state-of-the-art methods in both accuracy and profit maximization.
Prediction of bank telephone marketing results based on improved whale algorithms optimizing S_Kohonen network
The paper investigates a method to enhance the success rate of bank telemarketing for time deposits. It introduces an improved Whale Optimization Algorithm (WOA) to optimize an S_Kohonen network, which is a variation of the traditional Kohonen network with added supervised learning capabilities. This approach aims to improve the prediction accuracy of whether customers will subscribe to time deposit products based on telemarketing efforts. One cool thing from this paper is that it uses Chi-square and T-tests to determine which variables are worth using in the models.
How to improve the success of bank telemarketing? Prediction and interpretability analysis based on machine learning
The paper focuses on enhancing the effectiveness of telemarketing campaigns in banks using machine learning techniques. It develops and tests three machine learning models: Random Subspace (RS), Multi-Boosting (MB), and a combined Random Subspace-Multi-Boosting (RS-MB) model. Among these, the RS-MB model achieved the best performance in predicting the success of telemarketing calls.
Key findings of the paper include:
- Model Performance: The RS-MB model outperformed others, providing more accurate predictions.
- Variable Importance: The study identified critical variables influencing the success of telemarketing, such as the customer’s job type, the month of contact, and the day of the week.
- False Negatives vs. False Positives: The research emphasized the importance of reducing false negatives (missing potential customers) over false positives (contacting uninterested customers) for maximizing campaign success.
- Interpretability: Partial Dependence Plots (PDP) were used to show how important variables affect the likelihood of a customer subscribing to a term deposit, aiding in the interpretability of the models and providing actionable insights for bank marketing strategies.
That is all for today!
See you tomorrow :)