Project — Levi Jeans Site Analysis
Project — Levi Jeans Site Analysis
Brief
Levi Jeans
- Identify potential great opportunity sites for Levi Jeans.
- Identify variables for good and poor performance stores.
- Evaluate current Levi Jeans stores.
Tools to be used
SpatialXL - Spatial analytics in Excel
SpatialAL - Automated machine learning in Excel
RouteXL - Routing and logistics optimization
Geoscope - Spatial publisher to standalone viers files
Methodology
- Received all current Levi Jeans stores.
- Matched Levi Jeans stores to Shopping Centres – 16 stores didn't match to Shopping Centres.
- Using Shopping Centre classification we designed the territories of current and potential stores:
- Convenience, Neighbourhood, and Community – 1km drive distance territories and 3km in rural areas.
- Regional and Small Regional – 2km drive distance territories and 5km in rural areas.
- Super Regional – 8km drive distance territories.
- Using all client data provided plus all possible aggregated fields from Geo terra Image (GTI) datasets we populated each territory with the data.
- Then we used Automated Machine Learning (SpatialAL) as described below.
Automated Machine Learning
The profitability of a store is driven by a number of factors, some of them environmental. We don't have access to all the factors that lead to success or failure (for instance if the store is in a busy aisle of the shopping centre) but we put in all the factors that we do have access to (NLI, demographics, day-night population etc) and this is aggregated into the territory of each store (existing and new sites).
Using Machine Learning we take all the variables and let the computer analyse them and find patterns and see which variables influence profitability and calculate a fit for that. The more data we put into the Machine Learning algorithm the more accurate or better our fit is likely to be.
To test our fit, we run the current stores through the algorithm and test the variance (explained below). Once this test is successful, we apply this model to all the new store locations to predict profitability.
GTI Datasets
PrimeThought Software Solutions — includes Neighbourhood Lifestyle Index, Demographics, Advanced Population, Building Based Land Use, and New Developments data provided by our data partner Geo terra Image (GTI)
Neighbourhood Lifestyle Index (NLI)
- NLI classifications and LSM-to-NLI translation used for consumer segmentation.
Demographics
- Population and demographic data aggregated into each store territory.
Advanced Population
- Day-night population movement data for territory analysis.
Building Based Land Use
- Commercial and residential land use classification around store locations.
New Developments
- Tracking new developments that may impact store performance.
Methodology Continued
Using automated machine learning (SpatialAL – Spatial Automated Learning) we then predicted the current model's turnover by training the model with all variables and testing that it is correct. Using the verified model we then predicted turnover for all Shopping Centres where Levi Jeans stores are not present.
Results — Variance Report
The variance between predicted and actual will test whether you have relevant factors in your data. An average variance of 37% was obtained for predicting the turnover values with some higher outliers.
Top 10 Predicted Turnover — Levi Jeans Shopping Centre Territories
| Rank | Shopping Centre | Type | GLA | Territory | Predicted Turnover |
|---|---|---|---|---|---|
| 1 | Mutual Mall – Durban | Convenience | 5,000 | 1km Drive Distance, Urban | 970,394 |
| 2 | Pinecrest | Small Regional | 40,202 | 2km Drive Distance, Urban | 670,190 |
| 3 | Mega City | Small Regional | 47,076 | 5km Drive Distance, Rural | 668,371 |
| 4 | Gateway Theatre of Shopping | Super Regional | 158,319 | 8km Drive Distance | 666,704 |
| 5 | The Pavilion | Super Regional | 119,000 | 8km Drive Distance | 657,446 |
| 6 | Shoprite Centre – Ermelo | Neighbourhood | 7,023 | 1km Drive Distance, Urban | 655,749 |
| 7 | Musgrave Centre | Small Regional | 39,433 | 2km Drive Distance, Urban | 646,753 |
| 8 | Secunda Mall | Regional | 60,000 | 5km Drive Distance, Rural | 629,709 |
| 9 | Ermelo Mall | Community | 20,465 | 1km Drive Distance, Urban | 627,040 |
| 10 | Piet Retief SC | Neighbourhood | 7,436 | 1km Drive Distance, Urban | 576,419 |
Sample outputs for the top 10
