issues
search
mozhao0331
/
Restaurant_Segmentation_Analysis
MIT License
2
stars
2
forks
source link
2023-05-03 Official Partner Meeting Agenda
#11
Closed
erictsai1208
closed
1 year ago
erictsai1208
commented
1 year ago
Agenda
[x] Discuss our plan going forward with the project after yesterday's meeting.
[x] Discuss proposal: have the Sitewise team take a look at our presentation slide and proposal write-up if possible.
[x] Go through data dictionary and go through any abbreviations that are unclear.
Questions:
How were the 5 categories for Smoothie King derived?
Is the "mi" in terms of radius around a store location?
Is there any data involving the cost of leasing for an area?
Possible to get demo version of Sitewise app? (particularly the map that was shown during our first meeting)
Any suggestions on the best way to explore differences between categories for Smoothie King features?
CChCheChen
commented
1 year ago
Meeting Notes
Action Items
:
MDS team
: Current focus on EDA (mostly on categorical variable distributions over different categories for Smoothie King)
Sitewise team
: Extra variables (couple of hundreds for trade areas, 20-30 for comp/sister) will be delivered by end of the week
Sitewise team
: send over Sitewise app screenshot we need.
Meeting takeaways
:
Track mobile devices (geo fencing) over time (12 month) to determine what the category of Smoothie King is
Findings from Smoothie King can only be used as to provide domain knowledge which can help evaluate the model for Subway US.
Evaluate on restaurant revenue, not on profit
Try Recursive feature elimination (RFE) aside of PCA, but the ultimate goal is to find out the most important features
Smoothie King data 90/10 train test split. CV with 70/30.
Smoothie King Accuracy aim at 80% (20% error rate)
Other category: Church, Gym, Hospital and School Travel: Tourist and Hotel
Agenda
Questions: