Shop type
Linear Regression
Predicts profit/loss score as a weighted linear combination of all features. Fast, interpretable baseline model.
Random Forest
Simulates ensemble decision trees with noise reduction. Higher accuracy than single regression.
K-Means Clustering
Groups locations into High / Medium / Low potential clusters based on all feature distances.
Score comparison — Regression vs Random Forest
Side-by-side predicted scores per location. Random Forest (green) is more conservative and accurate.
RegressionRF
Linear Regression
Random Forest
Feature importance — Random Forest weights
How much each feature contributes to the RF prediction. Blue = helps score, Red = hurts score.
Linear Regression — predicted profit/loss by location
Profit/loss estimated using a linear weighted model. Green = profit, Red = loss.
Scatter — rent vs predicted score (Regression)
Relationship between rent cost and predicted business score. Identifies if high rent hurts profitability.
Profit
Marginal
Loss
Line chart — regression score trend across locations
Trend line of predicted scores sorted by rank. Shows how sharply scores drop from best to worst.
Random Forest — top locations by score
Horizontal bar ranking all locations by RF score. Color indicates K-Means cluster assignment.
Scatter — human traffic vs RF score
How foot traffic drives RF model output across locations. Color = cluster potential.
High potential
Medium potential
Low potential
Line chart — RF score vs competitor density
How competitor density impacts RF score. Sorted ascending by competitor density.
K-Means cluster plot — location groupings
Bubble plot: 3 clusters High/Medium/Low. Bubble size = nearby facilities. X = human traffic, Y = rent.
High potential
Medium potential
Low potential
Bubble size = nearby facilities
Cluster distribution
Count of locations per cluster.
Cluster score trend
RF scores across ranked locations, colored by cluster.
Bar chart — all locations by final score
Compare all locations at a glance. Color = K-Means cluster (green=high, amber=medium, red=low).
Heatmap — feature relationships across all locations
Green = favorable for business, yellow = neutral, red = unfavorable. Rent and competitor density are inverted (lower is better).
Scatter — rent vs score
Rent vs regression score trends and patterns.
Profit
Marginal
Loss
Cluster plot (K-Means output)
Grouped High/Medium/Low locations. Bubble size = nearby facilities.
Line chart — score trends across locations
Score trends for both ML models side by side. Useful for spotting model disagreements and outliers.
Linear Regression
Random Forest
Final recommendation table
Ranked by RF score. Cluster from K-Means. Outcome from Regression.
RegressionRFK-Means
| # | Location | Reg score | RF score | Cluster | Predicted profit | Human traffic | Vehicle traffic | Rent | Competitor density | Nearby facilities | Verdict |
|---|