Jordan 2.0 Study

Introduction

A new Jordan flagship store is about to be opened in Beijing Sanlitun area, and the purpose of this study is to help teams to have a better understanding on current landscape of Jordan related purchases in Bejing Urban area, to illustrate the customer flow within the Sanlitun commercial area, and to model how Beijing Live customers’ Jordan purchasing behavior.
This study focuses on four main aspect of the Jordan store:

Exploratory Data Analysis on orders and members within Beijing Urban area (study area).
Exploratory Data Analysis on orders and members of Beijing Live store. This store is also located in Sanlitun.
A binary model on the probability to buy a Jordan product for Beijing Live customers.
A geo-spatial analysis on the Location-based service data for Sanlitun commercial area.

Data

1. Online Transaction Data ¹

Online transaction data used includes all online transactions from all sources, including Nike.com, App, TMALL, etc. Time period is from 05/11/2020 to 06/20/2021.

2. Delivery Data ²

Delivery recipient’s district for each transaction is used to estimate customer’s active district. Data restricted to our study area.

3. Member Data ³

Member data includes information about the member who makes online purchases, including gender and age. Members who make both online purchases and BJ Live Inline purchase are isolated to study BJ Sanlitun Live.

4. Beijing Live inline Transaction Data ⁴

All transactions occured in Beijing Live since its opening.

5. Location-based Service Data ⁵

Location-based Service (LBS) is a software service which uses geographic data and information to provide services or information to users. The LBS data used includes customer flow, purchasing power, gender, etc.

6. Geo-spatial Data ⁶

A GeoJSON for geographic boundaries of all six districts is created, and a Gaode street map is used as the basemap. All coordinates are in GCJ09 system.

Study Area

Since the Beijing Sanlitun Area is a prime destination for Beijing, the study area specified is Beijing Urban area, which consist of six districts:

Dongcheng
Xicheng
Haidian
Chaoyang (where Sanlitun is located)
Shijingshan
Fengtai

Exploratory Data Analysis

1. Beijing Urban Area Online Orders

To explore the current demand of Jordan products at study area level, the online order information is grouped by Beijing urban district, using delivery address.

Beijing Urban Area Jordan Buying Member Count and Percentage for Online Orders

The interactive map shows that most Jordan Buying Members are in Chaoyang district, which is the district where Jordan 2.0 will be located, and all six districts have similar percentage of Jordan Buying Members. ⁷

Beijing Urban Area Online Order Category

Jordan products are the most popular products online in all districts. Along with basketball products are also very popular. There seems to be a high correlation between Jordan Brand products and basketball products.

2. Beijing Live Store

Beijing Live Store is located at Sanlitun South, which is the same commercial area as where Jordan will be located. To explore the inline purchase pattern of the customers from the study area, customers who purchased at Beijing Live Store are matched with their online delivery address, and their purchase category and member profile are aggregated to study area level. There are 2184 (n) customers who live in the study area and purchased at both Beijing Live Store and online.

Beijing Live Store ADPT and UPT

District	ADPT	UPT
Haidian	107	1.4
Chaoyang	103	1.5
Shijingshan	108	1.2
Xicheng	109	1.4
Fengtai	104	1.4
Dongcheng	113	1.4

Beijing Live Store Order Category

Customers who live in Chaoyang purchased a lot more products in Beijing Live comparing to other district. According to this chart, beside Nike Sportswear, Jordan products are very popular. From the previous charts, we suspect that there is a correlation between basketball product and Jordan product. We can model the relationship between both inline and online basketball product purchases and Jordan products.

Beijing Live Member by District

Most of the Beijing Live customers are from Chaoyang and Haidian.

Beijing Live Store Gender

Most of the Beijing Live customers are male. We will also model the relationship with gender to see if gender plays a role in Jordan purchase.

Beijing Live Store Age Group by Districts

Most of the customers are within 25-34 age group.

Model

The analysis in the previous chapter shows patterns and relationship in member and product category aspects, which indicates some possible variables to look at for modeling. Gender, age, online and inline purchase history are selected for further examination.

Gender and Age

The boxplot above shows that Male in general purchase more Jordan products in Live. Gender effect can be modeled to further explore the relationship.

Similar to the chart from previous chapter, Jordan product is purchased predominantly by customers age between 25-34. This may be due to live customer’s general age group is 25-34. The relationship between age and Jordan purchasing behavior needs to

Correlation - Purchase History

Correlation matrix confirmed that there is a strong correlation between Jordan and basketball products, for both online and BJ Live purchases.

Machine Learning - Binary Classification

Variables to be used in training:

Gender
Age Group
Online Purchase History - Basketball
Online Purchase History - Football
Online Purchase History - Jordan
Online Purchase History - Tennis
BJ Live Purchase History - Basketball
High Value Member - Member who purchased $375 or more in a year in the ecosystem

Dependent Variable: Buy Jordan or Not (Binary Variable) - processed from Online and Beijing Live purchase history

Algorithm used: Logistic Regression Classification

$Logit(P(Y = Buy Jordan(1))) = log(\frac{p}{1-p}) = \\ \beta_{0} + \beta_{1} Gender + \beta_{2}Age Group + \beta_{3} Basketball + \beta_{4} Football + \beta_{5} live \dots + \beta_{n}HighValue$

Logistic Regression is commonly used for binary classification problems. one advantage of this algorithm is that it gives $\beta$ for each term, and the signs can explain the effect of each term on dependent variable.

Model Result

Variable	Log Odds($\beta$)
Male	1.053
Basketball	1.305
Live_basketball	0.260
Live_football	-0.243
High_value_member	1.501

The model result confirms our understanding from the previous chapter. Male are more likely to purchase Jordan (a increase of 1.053 log odd comparing to female), and if the customer purchased basketball online or in BJ live, he or she is more likely to purchase Jordan. High Value Member is more likely to purchase Jordan. Age effect is statistically insignificant.

Model Performance

McFadden

The model fits pretty well, with a McFadden of 0.24, which is good for a logit regression.

Testing set (25% of the sample data) is used for measuring model’s prediction performance, and a ROC is created.

The model achieved an AUC of 0.80. The model can be used for both explanatory and predictive purposes.

Location-based Service

Flow Comparison North and South of Sanlitun

North Region: 0.256 Flow/M2 South Region: 0.676 Flow/M2

Sanlitun South region has a weekend flow of 26128 and Sanlitun north region has a weekend flow of 9262. South region’s flow is 2.6 times of the north region’s, adjusted by regions’ area. However, there is more male customers in north region.

Sanlitun Flow

This map shows the flow distribution of the Sanlitun commercial area. a clear flow line is shown from the south to north. The corridor connecting south and north region helps the customer to move quickly from south to north. Jordan 2.0 is located at the end of the main flow line, the entrance of the north region, and also the location with the highest flow and purchasing power (PP) in north region. Comparing to south region, north region has more male customers.

Conclusion

Beijing Urban Area has many potential of opportunities for Jordan. There is a high level of online demand for Jordan, and high level of basketball related purchase in Beijing Live. The model shows that the customer is more likely to purchase Jordan, if he or she purchase basketball related products in Beijing live Store. Gender (male) also play a huge role, and currently, there are more male customer in BJ Live, and LBS data shows that for Sanlitun Mall, there is more male customer than female. High value members are more likely to purchase Jordan products, and about 66% of the BJ Live members who live in the study area are high value members.

Acknowledgment

Thank Vivian Shi for the supervision and guidance.
Thank Mathilde Zhang for the assistance on processing some of the data.

Source: Nike Digital Order Line↩︎
Source: Baozun Delivery↩︎
Source: Nike Membership Data↩︎
Source: Nike Inline Order Line↩︎
Source: LBS Sample data↩︎
Source: Datav.GeoAtlas http://datav.aliyun.com/tools/atlas/index.html ↩︎
Jordan Buying Member $/$ All Members↩︎