Group07 Proposal

Overview

Proposal

Report

Poster

Application

All Projects

Project Background

E-commerce represented 54% of U.S. sales in the third quarter for TV and web merchant QVC.com. QVC is one of the most successful TV shopping broadcasters worldwide, which is one of eight leading retail brands under the Qurate Retail Group, focuses on building customer relationship, to engage with them to discover a dynamic catalogue of products from a wide range of categories. QVC is committed to providing its customers with thousands of the most innovative and contemporary beauty, fashion, jewellery and home products. Customers are engaged via multiple channels, through social media like Facebook Live, proprietary platforms that focuses on specific product types and video commerce based on QVC’s television channel.

Project Motivation

To conduct a better business strategy for e-commerce from unsupervised massive transaction data is extremely challenging. The best way to turn analytics, metrics, and raw data into a universal story that everyone can get on board with is through data visualization. With that, we can understand the impacts of a supply chain design change on service, sustainability and risk, as well as to understand our customer better. It unveils the abnormal pattern from the operational perspective, company would easily know where the pain points locate so as to make improvements, not only for the profitability of company itself, but also to provide customers with a more pleasant shopping experience. Because shopping is enjoying!

Project Objectives

The data from QVC, a giant Home-Shopping company, comprises millions of transactions and thousands of physical locations, traditional statistics figure can only tell partial story, our group decides to

• visualize the value behind data from a single Web-enabled tool
• reveal consumer behaviour patterns for the days of the week
• forecast the demand of certain products of the region from inventory management perspective.

We aim to generate a more efficient streamline upon current supply chain and provide an effortless too to help company make better decisions.

Project Scope

Our project consists of the following components:

1. Data cleaning and preparation: We need to remove invalid records from the dataset. We also need to prepare the dataset in a way that allows the plotting of network graphs. This enables users to see the flow of goods from warehouses to consumers, and could be helpful for distribution management.
2. Time series analysis and forecasting: Using historical data, we will identify if there is any seasonality in customer orders. Then, we will apply time-series forecasting methods to provide users an estimated demand figure for future periods, which could be helpful for inventory management and procurement planning.
3. Calendar plot: We will visualize the number of customer orders using a calendar plot. This enables users to see which months and days of week have higher demand, and this information could help them in manpower management. For example, if I know there are more demands coming in on weekends close to year-end, I can hire more temporary workers to process and deliver the orders.
4. Warehouse performance analysis: We will calculate and pinpoint the warehouses which have high proportions of delivery delays, and uncover possible reasons behind the delay. (e.g. certain product categories have more delayed deliveries; long distances between warehouse and consumer)
5. Distribution analysis: We will visualize if there are any geographical gaps between warehouse supply and consumer demand, which could help users decide if they should rearrange their stocks to more strategic locations.
6. Geospatial analysis: We may uncover spatial patterns in consumer demand, which can be helpful for distribution optimization by stocking spatially-correlated product categories or SKUs in warehouses of the same area.

Dataset Overview

Datasets are extracted from http://ibit.temple.edu/analytics/can-small-independent-pharmacies-compete-with-the-big-chains/
All 6 data sets with 36 columns were in CSV format, total number of rows after joining data sets - 4,680,635 rows. The final dataset covers 6 months time period [July 2016 - Dec 2016] and geographically represents U.S. territory.

Metadata

Variable	Description
Sales_Order_Nbr	Unique identifier for a given transaction
Sales_Order_Line_Nbr	Unique sub-identifier for a given transaction
Package_Id	ID code used to identify a specific box or package
Order_Dt	Date the transaction was placed
Party_Id	Unique identifier for an individual or organization
Order_Type_Cd	Code used to the type of Order
Shipping_Priority_Ind	Indicator used to indicate if order will be fulfilled with prioritized shipping (e.g. Overnight)
Total_Line_Amt	Total amount owed for the specific order line
Unit_Price_Amt	Merchandise price for the specific line
Line_Status_Cd	Status of the specific order line (refer to status reference sheet)
Line_Status_Dt	Date the specific order line was set to the current status
Product_Id	Unique identifier for an item presented for purchase (e.g. The number seen on TV)
Skn_Id	Internal ID used for back office purposes
Sku_Id	Unique ID used to denote a Product's specifics (e.g. specific color and size)
Color_Desc	Description of the color
Size_Desc	Description of the size
Shipped_Dt	Date the order line was shipped (left QVC's warehouse)
Source_Ship_Warehouse_Nbr	Unique identifier denoting the warehouse where the product is shipped from
Assigned_Dc_Id	Distribution center assigned to order at time of creation
Cancelled_Qty	Number of units cancelled
Ordered_Qty	Number of units requested by customer
Shipped_Qty	Number of units shipped to customer
Merchandise_Div_Desc	Description of Merchandise Division
Merchandise_Dept_Desc	Description of Merchandise Department
Carrier_Used_Tracking_Id	ID used to denote a unique package shipment
Shipment_Status_Dt	The most recent date of the shipment's status
Pickup_Dt	Date the package was picked up by the carrier from QVC
Scheduled_Delivery_Dt	Date the package is believed to arrive
Rescheduled_Delivery_Dt	Revised date the package is believed to arrive
Package_Scan_Dttm	Date the last time the package was scanned along its journey
Package_Cnt	Number of packages associated with a specific order line
Actual_Total_Package_Qty	Number of packages associated with a specific order line (post any consolidation)
Delivery_Confirmation_Dt	Date the package was delivered to the consumer
SHIP_TO_CITY	City the package was delivered to
SHIP_TO_STATE	State the package was delivered to
SHIP_TO_ZIP	Zip Code the package was delivered to

Deliverables

At the end of the project, we will submit the following deliverables:

• R Shiny web application
• Research paper
• Poster
• Project artifacts

Schedule

Project schedule