Sips and Stats: Seattle’s Best Coffee Spots Revealed

5 min read • June 18, 2025

Seattle’s coffee scene is iconic. But how does one cut through the hype to find the actual best spots? Atleast that is the problem I had in my initial days in the city and hence came the idea to explore the data for analysis.

YELP to the rescue! ;)
I used the Yelp API, unsupervised learning, geospatial clustering, and scaled review scores with some caffeine magic.

Data Collection

I pulled real-time data from Yelp’s Fusion API across Seattle, Bellevue, Kirkland, and Redmond. For each coffee shop, we extracted:

rating
review_count
price_level (converted from $ to numeric)
distance from city center
Geocoordinates (latitude, longitude)
Business metadata like categories and open/closed status

Choosing the Right Number of Clusters

Used the Elbow Method to find the optimal number of clusters (k) based on rating, reviews, price, and distance.

Questions

The elbow point clearly showed k = 4, which we used to form clusters with KMeans.

Clustering Seattle Coffee Shops

Each coffee shop was grouped into one of four clusters using:

rating
review_count
price_level
distance_km

These features were scaled using StandardScaler before clustering.

All clusters showed high average ratings (~4.0–4.7), but differed in price, location, and popularity. Something that justifies that coffee in Seattle has it’s own high benchmark standards :)

Questions

Map Visualization

We visualized the clusters on a folium map:

Seattle Coffee map

Each colored circle shows a coffee shop, color-coded by its assigned cluster.

Scaled Composite Score: Quality × Popularity

We created a scaled score to rank coffee shops by a balanced combination of quality and popularity.

from sklearn.preprocessing import MinMaxScaler

scaled = MinMaxScaler().fit_transform(df[['rating', 'review_count']])
df['scaled_rating'] = scaled[:, 0]
df['scaled_reviews'] = scaled[:, 1]
df['scaled_score'] = df['scaled_rating'] * df['scaled_reviews']

This avoids raw dominance of review counts and highlights well-loved shops across the board.

Top 10 Coffee Shops by Scaled Composite Score

Name	Location	Rating	Review Count	Scaled Score
Biscuit Bitch	1909 1st Ave	4.2	5030	0.840000
Starbucks Reserve Roastery Seattle	1124 Pike St	4.4	4530	0.792525
The Crumpet Shop	1503 1st Ave	4.5	2707	0.484354
Storyville Coffee Company	94 Pike St	4.6	2474	0.452501
Top Pot Doughnuts	2124 5th Ave	4.0	2048	0.325726
Biscuit Bitch	2303 3rd Ave	4.3	1829	0.312712
Moore Coffee Shop	1930 2nd Ave	4.4	1335	0.233559
Anchorhead Coffee - CenturyLink Plaza	1600 7th Ave	4.5	983	0.175885
Coffeeholic House	3700 S Hudson St	4.7	782	0.146139
Espresso Vivace Roasteria	532 Broadway Ave E	4.1	874	0.142481

💸 Best Value-for-Money Coffee Spots

To surface affordable gems, we also calculated:

value_score = rating / price_level

This helps highlight high-rated spots with lower price points.

🔟 Top 10 Value-for-Money Coffee Shops

Name	Rating	Price	Price Level	Value Score
Overcast Coffee Company	4.8	$	1.0	4.8
URL Coffee	4.7	$	1.0	4.7
The Bridge Coffee House	4.7	$	1.0	4.7
Sound & Fog	4.7	$	1.0	4.7
Experience Tea	4.7	$	1.0	4.7
C.C Espresso and Ice Creamery	4.7	$	1.0	4.7
Cafe Argento	4.6	$	1.0	4.6
Squirrel Chops	4.6	$	1.0	4.6
Phin	4.6	$	1.0	4.6
The Station	4.6	$	1.0	4.6

🧠 What We Learned

Seattle’s coffee scene is remarkably high-rated across the board
Review count isn’t always correlated with value
Clustering helps organize the scene; scoring helps spotlight the stars

🚀 Next Steps

Use time-decay scoring for recent reviews
Sentiment analysis from textual reviews
Cluster-specific rankings (e.g., best hidden gem)

Explore the code and try your own ranking here:
🔗 Open the Colab Notebook

If you enjoyed this, follow Vishnu on LinkedIn or check out more blogs at mlgyaan.com for practical machine learning, data science, and storytelling with code.