Restaurant Review Data Analysis and Visualization with Python and pyecharts

When a friend needed a data‑driven marketing plan for the restaurant industry, the author suggested starting with online reviews from platforms such as Meituan, Dianping, and Ele.me, and then built a Python demo to illustrate the workflow.

The demo begins by fetching a sample of Dianping reviews via an API, anonymising the store name and address, and converting the JavaScript‑style literals true/false/null to Python's True/False/None. The cleaned JSON is saved as review_data in a data.py file for easy import.

A second script analysis.py imports the data and verifies its structure. From the summarys list inside the data, the overall impression scores are extracted and visualised with a customised pyecharts bar chart. The X‑axis labels (e.g., "菜品健康", "牛肉赞", "回头客") and Y‑axis counts (e.g., 51, 32, 29) are obtained via list comprehensions:

summary_name_list = [i.get('summaryString') for i in summarys]
summary_count_list = [i.get('summaryCount') for i in summarys]

The bar chart’s add_yaxis() call uses category_gap=80% to increase column spacing for clarity.

Next, the tutorial extracts individual star ratings (1–5 stars, represented as 10, 20, 30, 40, 50) from each review, builds a list, and applies collections.Counter to count occurrences:

all_star = [i.get('reviewDataVO').get('reviewData').get('star') for i in all_review]
stars = dict(Counter(all_star))

The resulting dictionary, e.g., {30: 2, 50: 3, 10: 1, 20: 3, 35: 1} , provides the data for a pyecharts pie chart after converting keys and values to lists:

data = zip(list(stars.keys()), list(stars.values()))

Both the bar and pie charts are rendered, showing that the sample restaurant receives mixed feedback, with a notable proportion of 2‑ and 3‑star ratings and relatively few 5‑star reviews.

Finally, the author combines the four generated charts into a single report, demonstrating how Python’s built‑in tools (like Counter) and the pyecharts library can be leveraged for quick, insightful data analysis that can evolve into a full commercial solution.