-
Used RFM method to calculate Recency, Frequency, Monetary to make a segmentation based on customer habits
-
Receny : How recently the customer made a transaction.
-
Frequency : How often the customer make transactions.
-
Monetary : How many transactions the customer has made.
-
Used Quartile statistical method to calculate the Individual RFM score
-
we have assigned scores from 1 to 4.
-
4 being the highest and best, while 1 being the lowest.
-
Used K-Means Clustering unsupervised Technique to cluster the customers depend on their habits
-
K-Means Clustering is an unsupevised machine learning algorithm that uses multiple iterations to segment the unlabeled data points into k different clusters.
-
For k_means clustering to give best result the followings conditions should be met:
- Data distribution must not be skewed.
- Data is standardised
- used Elbow method to find the optimum number of clusters.
- Then used Davies Bouldin and silhouetter score to evaluate the model and optimise the k value.