Advanced Analytics for Customer Retention
Budget: $10 – $30 USD
Summary
A retail company aims to improve customer retention by applying advanced analytics techniques. The company has historical data for 200 customers, including observed churn risk, and data for 100 new customers for whom churn is unknown. Your consultancy team is tasked with developing an analytics-driven decision support framework that integrates predictive modelling, customer segmentation, and multi-criteria decision-making.
Dataset Description
The dataset contains 300 customers and the following variables: - Customer_ID - Age - Annual Income (EUR) - Average Monthly Spend (EUR) - Purchase Frequency per Year - Website Visits per Month - Churn Risk Percentage (available for 200 customers only) .
Data set excel sheet will be provided.
Task 1: Churn Prediction Using ANFIS
Using the 200 customers with known churn risk, develop an ANFIS model to predict customer churn probability. The model should use the available demographic and behavioural variables as inputs. Validate the model using appropriate performance metrics and then use it to predict churn probability for the remaining 100 new customers.
Task 2: Customer Clustering
Normalise the dataset and apply a clustering technique (Self-Organising Map or K-means) to segment all 300 customers into exactly nine (9) clusters. The clustering should be based on behavioural variables and predicted churn risk. Clearly justify the choice of method and parameters.
Task 3: Cluster Ranking Using MCDM
Treat the nine clusters as decision alternatives and rank them using a Multi-Criteria Decision-Making (MCDM) techniques. Justify the selection of criteria from the dataset and weights and interpret the final ranking from a managerial perspective.
Report Requirements
The report (2,000 words) should include:
1. Introduction and problem formulation
2. Data understanding and preprocessing
3. ANFIS model development and results
4. Clustering analysis and cluster interpretation
5. MCDM application and cluster ranking
6. Managerial insights and conclusions
Submission Instructions
Must submit:
1. A single word format report
2. The analysis file(s) used to conduct the modelling (Excel)
Important aspects to consider when writing report.
• Churn prediction and ANFIS modelling
• Clustering and data normalisation
• MCDM ranking and interpretation
• managerial insight
A retail company aims to improve customer retention by applying advanced analytics techniques. The company has historical data for 200 customers, including observed churn risk, and data for 100 new customers for whom churn is unknown. Your consultancy team is tasked with developing an analytics-driven decision support framework that integrates predictive modelling, customer segmentation, and multi-criteria decision-making.
Dataset Description
The dataset contains 300 customers and the following variables: - Customer_ID - Age - Annual Income (EUR) - Average Monthly Spend (EUR) - Purchase Frequency per Year - Website Visits per Month - Churn Risk Percentage (available for 200 customers only) .
Data set excel sheet will be provided.
Task 1: Churn Prediction Using ANFIS
Using the 200 customers with known churn risk, develop an ANFIS model to predict customer churn probability. The model should use the available demographic and behavioural variables as inputs. Validate the model using appropriate performance metrics and then use it to predict churn probability for the remaining 100 new customers.
Task 2: Customer Clustering
Normalise the dataset and apply a clustering technique (Self-Organising Map or K-means) to segment all 300 customers into exactly nine (9) clusters. The clustering should be based on behavioural variables and predicted churn risk. Clearly justify the choice of method and parameters.
Task 3: Cluster Ranking Using MCDM
Treat the nine clusters as decision alternatives and rank them using a Multi-Criteria Decision-Making (MCDM) techniques. Justify the selection of criteria from the dataset and weights and interpret the final ranking from a managerial perspective.
Report Requirements
The report (2,000 words) should include:
1. Introduction and problem formulation
2. Data understanding and preprocessing
3. ANFIS model development and results
4. Clustering analysis and cluster interpretation
5. MCDM application and cluster ranking
6. Managerial insights and conclusions
Submission Instructions
Must submit:
1. A single word format report
2. The analysis file(s) used to conduct the modelling (Excel)
Important aspects to consider when writing report.
• Churn prediction and ANFIS modelling
• Clustering and data normalisation
• MCDM ranking and interpretation
• managerial insight