DATA ANALYTICS PROJECTS

OVERVIEW

This project is one of many projects provided by DQlab. It is about analyzing, giving the business insight, and building the model of sales company data from 2013 - 2019. The company really wants to know how many customers with its status 'Churn', how it gives the effect of transaction value, and build the model to detect them automatically.

The data provided has a total data of 100,000 samples. The process starts from data cleaning, feature engineering, data visualization, and model training. In the feature engineering process, I defined a churned customer as a customer who did not make a transaction less than August 1, 2018. Then, I also did binning for the Average Transaction Column and the Transaction Count Column.

For better visualization, I turned all the charts into one simple dashboard report with Tableau software. It is shown in the image below. In the process of training the model, I use the Support Vector Classifier algorithm with a training set proportion and validation of 75:25. The results of the model training show that the level of accuracy obtained is 0.77.

SUMMARY

• The total transaction value has consistently increased from 2013 to 2015 in line with the increase in the number of customers. However, until 2018, although the number of customers continued to increase, the total transaction value tended to stagnate and even decreased drastically in 2018. In that year, the number of customers reached the highest point in the last 5 years but on the Contrary to the total transaction value drop drastically. It means that there are many customers who have made transactions before but are currently no longer actively made transaction, or they are referred to as Churn customers, starting in 2018
• The most popular products are shoes and jackets. Meanwhile, clothing products are the products with the highest average transaction value compared to other products and also with the least number of enthusiasts. Jacket products are the most promising products because they tend to increase in total transaction value while clothes experience a decrease. Shoe products are also quite promising as an alternative product choice
• There are about 60% of current customers with 'Churn' status in all product categories.
• The average value of transactions made by the most customers is in the price range of Rp. 1,000,000.00 - Rp. 2,500,000.00.

RECOMMENDATION

The company has to focus on promoting the most profitable product to the especially the churn customers with certain marketing strategy. There are two products that are quite promising, jacket and shoes. The company should give the attractive and tempting promotions to them so they will be interested in making transactions again. Just focus on those two products first.

‍