This repository holding a case study on analysis churn in Telecom and building a machine learning model to classify the customer who is likely to churn, which to improved customer satisfaction and reduced the churn rate by identifying customers who are likely to churn and finding insights from the data. Comes with a machine learning model for predicting the probability of customers' churn with Flask API, and visualized with Power BI Dashboard.
The dataset used in this project is from IBM Sample Data Sets, which hosted on Kaggle. For more information, please refers to the Kaggle dataset description.
[video on application]
| Document | Progress | Version | Links | |
|---|---|---|---|---|
| 1 | Explosive Data Analysis | DONE | 1 | Telecom_churn_EDA .ipynb |
| 2 | Model Selection with lazypredict | DONE | 1 | Telecom_churn_Model_Building_(Lazypredict).ipynb |
| 3 | Model Building | DONE | 1 | Telecom_churn_Model_Building.ipynb |
| 4 | Model Deployment with FLASK | DONE | 1 | Local Deploy: 1. html format 2. Main python files 3. Built Model |
| 5 | Model Deployment with FLASK on Google Cloud | ON-GOING | ||
| 6 | Presentation Notes | DONE | 1 | ML - Churn.pdf |
| 7 | Power BI Dashboard | DONE | 1 | Power BI repository |
【顧客流失預測項目】1. 數據說的故事要好好聽
【顧客流失預測項目】2. 模型會長怎樣
【顧客流失預測項目】3. 齊齊預測最快樂 - 將模型變成一個Web應用