What Customers Say — NLP Dashboard

Phrase Cloud

Most Distinctive Phrases by Rating

Click a star rating — the cloud rebuilds instantly showing phrases most exclusive to that rating. Size = frequency · Opacity = exclusivity vs. all other ratings.

Rating

Distributions

Rating Distributions

All dimensions skew toward 5-star — but the spread differs meaningfully across dimensions.

Topic Fingerprint

Topic Profile by Star Rating

Each rating has a distinct topic fingerprint. Toggle ratings to compare — the shape shifts dramatically between 1-star and 5-star.

Compare

T1 (Wait Time) spikes to 23.7% in 1-star — 3× higher than 5-star. T7 (Excellence) dominates 5-star at 19.3% — nearly 4× its 1-star presence.

Correlations

Rating Dimension Correlations

How strongly does each dimension predict Overall? Hover any cell for a plain-English interpretation.

Food → Overall r=0.90 is the strongest driver. Atmosphere → Overall r=0.77 is weakest — a beautiful room alone cannot save a bad meal.

Joint Calculator

The 95.9% Rule — When Does Overall Hit 5-Stars?

Each cell = % of reviews giving Overall=5 when Food (rows) and Service (columns) have those values. Hover any cell for review count and interpretation. The pattern is unmistakable.

When Food=5 AND Service=5 (43.4% of all reviews), Overall is also 5-star in 95.9% of cases. Drop either to 4 and the probability falls to ~73%. Drop to 3 and it collapses below 20%.

Topic Heatmap

Topic Prevalence Across All Ratings — Full Picture

Mean proportion of each topic in reviews at each star level. Click any bar for topic details. Green = dominant · Red = rare.

1-star DNA: T1 Wait Time (23.7%) + T9 Food Issues (17.6%) + T4 Mixed Experiences (14.4%) · 5-star DNA: T7 Excellence (19.3%) + T2 Recommendations (13.4%) + T5 Friendly Staff (12.4%)