Capstone Project Marketing and Retail Analytics
Budget: ₹600 – ₹1,500 INR
Now, OList is one such e-commerce company that has faced some losses recently and they want to manage their inventory very well so as to reduce any unnecessary costs that they might be bearing. In this assignment, you have to manage the inventory cost of this e-commerce company OList. You need to identify top products that contribute to the revenue and also use market basket analysis to analyse the purchase behaviour of individual customers to estimate with relative certainty, what items are more likely to be purchased individually or in combination with some other products.
You saw a detailed description of what each column signifies in the dataset. It is important to have a data dictionary to get a domain-based perspective of the database and problem statement. You also learnt how Shreyas looks at the possible values that a column can have to understand its data type and the range of values. Some things to keep in mind are:
For this capstone, consider only the cases having order status as 'delivered'.
Perform EDA before inferring anything from the dataset and treat the outliers and missing values carefully.After developing an understanding of the problem statement and the dataset, it is important to think about your approach to solve the problem. It is always advisable to break down a complex problem into smaller, simpler ones. This is called structured problem solving
As mentioned by Shreyas, the structured problem approach that you will be using here is the 80-20 rule of the Pareto analysis. In general, this 80-20 rule asserts that 80% of outcomes (or outputs) result from 20% of all causes (or inputs) for any given event. Again, the 80-20 number is not iron-clad and can be tweaked a little on a case by case basis
After using Pareto for your analysis let us see what all other important aspects are there that you need to consider to solve the problem
What should be the ideal category depth?
Some product combinations are sold more than others. This needs to be understood using market basket analysis.
After having specific categories in hand that will be beneficial to the company, how can you reduce the number of items in a particular category?
After gaining insights from the dataset, you will need to present your solution along with visualisation in the form of a dashboard and supplement it with an executive summary. This will help the team at OList to take the required action based on your recommendations.
Now that you have understood the various aspects that need to be taken care of while solving this capstone project, let's quickly summarise it all and take a look at the project pipeline you need to follow in order to come up with a fruitful analysis and end result.
Project pipeline
Data exploration and cleaning: Identify missing and duplicate values in each column and treat them accordingly. Also, treat any and all data quality issues associated with the dataset.
Data visualisation: Create appropriate visualisations to identify the most ordered products by quantity and revenue.
Market basket analysis: Identify combinations of product categories that are ordered frequently and visualise them as required. This can be done by creating a simple matrix with product categories in the rows and columns and then checking which of the pairs show associations.
Dashboarding and final PPT: Create a neat dashboard containing the important visualisations and supplement it with an executive summary containing valuable insights and recommended steps of action.
Submissions required
You will need to submit the following documents for this capstone project:
A clean version of the retail dataset provided at the beginning of this session.
A file containing a dashboard with all the important visualisations used in this project.
A PPT file with an executive summary containing insights and recommended steps of action to be presented to a business stakeholder.
A video explaining the presentation.
You saw a detailed description of what each column signifies in the dataset. It is important to have a data dictionary to get a domain-based perspective of the database and problem statement. You also learnt how Shreyas looks at the possible values that a column can have to understand its data type and the range of values. Some things to keep in mind are:
For this capstone, consider only the cases having order status as 'delivered'.
Perform EDA before inferring anything from the dataset and treat the outliers and missing values carefully.After developing an understanding of the problem statement and the dataset, it is important to think about your approach to solve the problem. It is always advisable to break down a complex problem into smaller, simpler ones. This is called structured problem solving
As mentioned by Shreyas, the structured problem approach that you will be using here is the 80-20 rule of the Pareto analysis. In general, this 80-20 rule asserts that 80% of outcomes (or outputs) result from 20% of all causes (or inputs) for any given event. Again, the 80-20 number is not iron-clad and can be tweaked a little on a case by case basis
After using Pareto for your analysis let us see what all other important aspects are there that you need to consider to solve the problem
What should be the ideal category depth?
Some product combinations are sold more than others. This needs to be understood using market basket analysis.
After having specific categories in hand that will be beneficial to the company, how can you reduce the number of items in a particular category?
After gaining insights from the dataset, you will need to present your solution along with visualisation in the form of a dashboard and supplement it with an executive summary. This will help the team at OList to take the required action based on your recommendations.
Now that you have understood the various aspects that need to be taken care of while solving this capstone project, let's quickly summarise it all and take a look at the project pipeline you need to follow in order to come up with a fruitful analysis and end result.
Project pipeline
Data exploration and cleaning: Identify missing and duplicate values in each column and treat them accordingly. Also, treat any and all data quality issues associated with the dataset.
Data visualisation: Create appropriate visualisations to identify the most ordered products by quantity and revenue.
Market basket analysis: Identify combinations of product categories that are ordered frequently and visualise them as required. This can be done by creating a simple matrix with product categories in the rows and columns and then checking which of the pairs show associations.
Dashboarding and final PPT: Create a neat dashboard containing the important visualisations and supplement it with an executive summary containing valuable insights and recommended steps of action.
Submissions required
You will need to submit the following documents for this capstone project:
A clean version of the retail dataset provided at the beginning of this session.
A file containing a dashboard with all the important visualisations used in this project.
A PPT file with an executive summary containing insights and recommended steps of action to be presented to a business stakeholder.
A video explaining the presentation.