create-viz
用 Python 把資料變成可出版品質的圖表，並挑對圖表類型
中文翻譯
---
name: create-viz
description: 用 Python 製作可出版品質的視覺化圖表。用在把查詢結果或 DataFrame 變成圖表、為趨勢或比較挑對圖表類型、為報告或簡報產生圖，或需要有 hover 和縮放的互動式圖表時。
argument-hint: "<data source> [chart type]"
---
# /create-viz - 建立視覺化圖表
> 如果看到不熟的佔位符，或需要確認目前連了哪些工具，請看 [CONNECTORS.md](../../CONNECTORS.md)。
用 Python 製作可出版品質的資料視覺化。依據資料產生圖表，遵循清晰、準確和設計上的最佳實務。
## 用法
```
/create-viz <data source> [chart type] [additional instructions]
```
## 工作流程
### 1. 了解需求
要判斷：
- **資料來源**：查詢結果、貼上的資料、CSV/Excel 檔，或需要去查詢的資料
- **圖表類型**：使用者明確指定的，或需要推薦的
- **目的**：探索、簡報、報告、儀表板元件
- **受眾**：技術團隊、高階主管、外部利害關係人
### 2. 取得資料
**如果有連接資料倉儲，而且資料需要查詢：**
1. 寫好並執行查詢
2. 把結果載入 pandas DataFrame
**如果資料是貼上或上傳的：**
1. 把資料解析成 pandas DataFrame
2. 視需要清理和準備（型別轉換、處理空值）
**如果資料來自對話中先前的分析：**
1. 直接參照現有的資料
### 3. 選擇圖表類型
如果使用者沒指定圖表類型，就根據資料和問題推薦一種：
| 資料關係 | 建議圖表 |
|---|---|
| 隨時間的趨勢 | 折線圖 |
| 跨類別比較 | 長條圖（類別多時用橫向） |
| 部分對整體的組成 | 堆疊長條圖或面積圖（除非少於 6 個類別，否則避免圓餅圖） |
| 數值的分布 | 直方圖或箱形圖 |
| 兩個變數的相關性 | 散佈圖 |
| 兩個變數隨時間的比較 | 雙軸折線圖或分組長條圖 |
| 地理資料 | 分級統計圖（choropleth） |
| 排名 | 橫向長條圖 |
| 流向或流程 | 桑基圖（Sankey） |
| 關係矩陣 | 熱度圖 |
如果使用者沒有指定，就簡短說明推薦的理由。
### 4. 產生視覺化圖表
依需求，用下列其中一個函式庫寫 Python 程式碼：
- **matplotlib + seaborn**：最適合靜態、可出版品質的圖表。預設選擇。
- **plotly**：最適合互動式圖表，或使用者要求互動性的時候。
**程式碼要求：**
```python
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
# 設定專業的樣式
plt.style.use('seaborn-v0_8-whitegrid')
sns.set_palette("husl")
# 建立大小適當的圖
fig, ax = plt.subplots(figsize=(10, 6))
# [圖表專屬的程式碼]
# 一定要包含：
ax.set_title('Clear, Descriptive Title', fontsize=14, fontweight='bold')
ax.set_xlabel('X-Axis Label', fontsize=11)
ax.set_ylabel('Y-Axis Label', fontsize=11)
# 適當地格式化數字
# - 百分比：'45.2%'，不是 '0.452'
# - 貨幣：'$1.2M'，不是 '1200000'
# - 大數字：'2.3K' 或 '1.5M'，不是 '2300' 或 '1500000'
# 移除多餘的圖表雜訊
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.tight_layout()
plt.savefig('chart_name.png', dpi=150, bbox_inches='tight')
plt.show()
```
### 5. 套用設計最佳實務
**顏色：**
- 使用一致、對色盲友善的調色盤
- 顏色要有意義（不是裝飾用）
- 用對比色突顯關鍵的資料點或趨勢
- 把比較不重要的參考資料調成灰色
**字體排印：**
- 標題要寫出洞察，而不只是指標（例如「營收年增 23%」，而不是「每月營收」）
- 軸標籤要易讀（盡量不要旋轉 90 度）
- 在關鍵點加上資料標籤，只要能讓圖更清楚
**版面：**
- 適當的留白和邊距
- 圖例位置不要遮住資料
- 類別依數值排序（不要依字母排序），除非有自然的順序
**準確性：**
- 長條圖的 Y 軸要從零開始
- 沒有清楚標註的話，不要用會誤導的軸截斷
- 比較多個面板時要用一致的刻度
- 適當的精確度（不要顯示 10 位小數）
### 6. 儲存與呈現
1. 把圖表存成 PNG 檔，取一個有描述性的檔名
2. 把圖表顯示給使用者看
3. 附上用到的程式碼，讓他們可以修改
4. 建議變化版本（不同的圖表類型、不同的分組、放大的時間範圍）
## 範例
```
/create-viz Show monthly revenue for the last 12 months as a line chart with the trend highlighted
```
```
/create-viz Here's our NPS data by product: [pastes data]. Create a horizontal bar chart ranking products by score.
```
```
/create-viz Query the orders table and create a heatmap of order volume by day-of-week and hour
```
## 小提醒
- 想要互動式圖表（hover、縮放、篩選），就提到「interactive」，Claude 會改用 plotly
- 需要較大的字體和較高的對比，就指定「presentation」
- 可以一次要求多張圖表（例如「create a 2x2 grid of charts showing...」）
- 圖表會以 PNG 檔存到你目前的目錄英文原文
---
name: create-viz
description: Create publication-quality visualizations with Python. Use when turning query results or a DataFrame into a chart, selecting the right chart type for a trend or comparison, generating a plot for a report or presentation, or needing an interactive chart with hover and zoom.
argument-hint: "<data source> [chart type]"
---
# /create-viz - Create Visualizations
> If you see unfamiliar placeholders or need to check which tools are connected, see [CONNECTORS.md](../../CONNECTORS.md).
Create publication-quality data visualizations using Python. Generates charts from data with best practices for clarity, accuracy, and design.
## Usage
```
/create-viz <data source> [chart type] [additional instructions]
```
## Workflow
### 1. Understand the Request
Determine:
- **Data source**: Query results, pasted data, CSV/Excel file, or data to be queried
- **Chart type**: Explicitly requested or needs to be recommended
- **Purpose**: Exploration, presentation, report, dashboard component
- **Audience**: Technical team, executives, external stakeholders
### 2. Get the Data
**If data warehouse is connected and data needs querying:**
1. Write and execute the query
2. Load results into a pandas DataFrame
**If data is pasted or uploaded:**
1. Parse the data into a pandas DataFrame
2. Clean and prepare as needed (type conversions, null handling)
**If data is from a previous analysis in the conversation:**
1. Reference the existing data
### 3. Select Chart Type
If the user didn't specify a chart type, recommend one based on the data and question:
| Data Relationship | Recommended Chart |
|---|---|
| Trend over time | Line chart |
| Comparison across categories | Bar chart (horizontal if many categories) |
| Part-to-whole composition | Stacked bar or area chart (avoid pie charts unless <6 categories) |
| Distribution of values | Histogram or box plot |
| Correlation between two variables | Scatter plot |
| Two-variable comparison over time | Dual-axis line or grouped bar |
| Geographic data | Choropleth map |
| Ranking | Horizontal bar chart |
| Flow or process | Sankey diagram |
| Matrix of relationships | Heatmap |
Explain the recommendation briefly if the user didn't specify.
### 4. Generate the Visualization
Write Python code using one of these libraries based on the need:
- **matplotlib + seaborn**: Best for static, publication-quality charts. Default choice.
- **plotly**: Best for interactive charts or when the user requests interactivity.
**Code requirements:**
```python
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
# Set professional style
plt.style.use('seaborn-v0_8-whitegrid')
sns.set_palette("husl")
# Create figure with appropriate size
fig, ax = plt.subplots(figsize=(10, 6))
# [chart-specific code]
# Always include:
ax.set_title('Clear, Descriptive Title', fontsize=14, fontweight='bold')
ax.set_xlabel('X-Axis Label', fontsize=11)
ax.set_ylabel('Y-Axis Label', fontsize=11)
# Format numbers appropriately
# - Percentages: '45.2%' not '0.452'
# - Currency: '$1.2M' not '1200000'
# - Large numbers: '2.3K' or '1.5M' not '2300' or '1500000'
# Remove chart junk
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.tight_layout()
plt.savefig('chart_name.png', dpi=150, bbox_inches='tight')
plt.show()
```
### 5. Apply Design Best Practices
**Color:**
- Use a consistent, colorblind-friendly palette
- Use color meaningfully (not decoratively)
- Highlight the key data point or trend with a contrasting color
- Grey out less important reference data
**Typography:**
- Descriptive title that states the insight, not just the metric (e.g., "Revenue grew 23% YoY" not "Revenue by Month")
- Readable axis labels (not rotated 90 degrees if avoidable)
- Data labels on key points when they add clarity
**Layout:**
- Appropriate whitespace and margins
- Legend placement that doesn't obscure data
- Sorted categories by value (not alphabetically) unless there's a natural order
**Accuracy:**
- Y-axis starts at zero for bar charts
- No misleading axis breaks without clear notation
- Consistent scales when comparing panels
- Appropriate precision (don't show 10 decimal places)
### 6. Save and Present
1. Save the chart as a PNG file with descriptive name
2. Display the chart to the user
3. Provide the code used so they can modify it
4. Suggest variations (different chart type, different grouping, zoomed time range)
## Examples
```
/create-viz Show monthly revenue for the last 12 months as a line chart with the trend highlighted
```
```
/create-viz Here's our NPS data by product: [pastes data]. Create a horizontal bar chart ranking products by score.
```
```
/create-viz Query the orders table and create a heatmap of order volume by day-of-week and hour
```
## Tips
- If you want interactive charts (hover, zoom, filter), mention "interactive" and Claude will use plotly
- Specify "presentation" if you need larger fonts and higher contrast
- You can request multiple charts at once (e.g., "create a 2x2 grid of charts showing...")
- Charts are saved to your current directory as PNG files