User question
    ↓
Agent understands the user's request
    ↓
Filter-based data lookup through the Excel connector (List rows present in a table)
    ↓
Filtered data → passed to the AI Prompt tool
    ↓
Customized analysis/HTML response generated

An ODATA filter query to restrict the entries returned.
Target lookup column: select either [YearMonth | District].
Target lookup key value: the input value changes depending on the key column.
YearMonth is in YYYYMM format.
For District, select one of the following: [All, Gangnam-gu, Gangdong-gu, Gangbuk-gu, Gangseo-gu, Gwanak-gu, Gwangjin-gu, Guro-gu, Geumcheon-gu, Nowon-gu, Dobong-gu, Dongdaemun-gu, Dongjak-gu, Mapo-gu, Seodaemun-gu, Seocho-gu, Seongdong-gu, Seongbuk-gu, Songpa-gu, Yangcheon-gu, Yeongdeungpo-gu, Yongsan-gu, Eunpyeong-gu, Jongno-gu, Jung-gu, Jungnang-gu]

Compare electricity usage in Dongdaemun-gu and Seoul.

Time Series Data Analysis Prompt

Choose freely

In the model response pane, change the output on the right from default text to JSON.

You are a time-series data analysis expert.
Based on the raw data imported from the Excel table, extract data insights that match the user's request.

Create two outputs: a text format that the user can read immediately,
and an HTML format that can be sent directly by email.
Respond in JSON format.



## Input
- User request: [Request]
- Time-series raw data: [RAW_Data]

## Analysis rules
1. Understand the user's request first, and select only the data rows related to the request.
2. From the selected data, derive insights from the following perspectives:
   - Sum, average, maximum/minimum values and the corresponding points in time
   - Trend judgment (increase/decrease/stable)
   - Identify outliers (values more than ±30% away from the average)
   - Key patterns (by day of week, month, season, etc.)
3. Never guess or generate values that do not exist in the data.
4. Include thousands separators in numbers and clearly indicate units.

## Output format
Respond using the structure below.

### 📊 Analysis overview
- Analysis target: [data range or topic]
- Analysis period: [start date ~ end date]
- Number of data records: [number of selected rows]

### 📋 Key summary
| Metric | Value |
|------|-----|
| Sum | [value + unit] |
| Average | [value + unit] |
| Maximum | [value + unit] ([point in time]) |
| Minimum | [value + unit] ([point in time]) |

### 💡 Insights
Each insight must be written in the following format:

**[Insight 1 title]**
- Content: [description of analysis result]
- Supporting data: [specific data values and points in time used to derive this insight]

**[Insight 2 title]**
- Content: [description of analysis result]
- Supporting data: [specific data values and points in time used to derive this insight]

(Write only as many insights as the data supports)

### 📈 Visualization (if applicable)
If there is a chart generated by the code interpreter, include it here.

### ⚠️ Outliers / notable items
- [outlier value], [point in time] — [deviation % from average]

### 🔍 Supporting data details
Summarize the key data used in the analysis as a table:

| [time column] | [value column] | Notes |
|-----------|---------|------|
| [point in time] | [value] | [above/below average/outlier, etc.] |

## Notes
- If the user specifies a specific region, period, or condition, analyze only that range.
- If there is not enough data to perform a reliable analysis, state that limitation.
- If the supporting data details table exceeds 20 rows, show only the key top/bottom items and summarize the rest by count and average.

Compare and analyze electricity usage for all of Seoul and Dongdaemun-gu.

[
  {
    "YearMonth": "202301",
    "City/Province": "Seoul",
    "District": "Dongdaemun-gu",
    "Households covered": "194,302",
    "Average household electricity usage (kWh)": "232",
    "Average household electricity bill (KRW)": "30,224"
  },
  {
    "YearMonth": "202308",
    "City/Province": "Seoul",
    "District": "Dongdaemun-gu",
    "Households covered": "196,419",
    "Average household electricity usage (kWh)": "312",
    "Average household electricity bill (KRW)": "46,034"
  },
  {
    "YearMonth": "202508",
    "City/Province": "Seoul",
    "District": "Dongdaemun-gu",
    "Households covered": "203,974",
    "Average household electricity usage (kWh)": "344",
    "Average household electricity bill (KRW)": "53,497"
  },
  {
    "YearMonth": "202301",
    "City/Province": "Seoul",
    "District": "All",
    "Households covered": "4,807,229",
    "Average household electricity usage (kWh)": "247",
    "Average household electricity bill (KRW)": "33,227"
  },
  {
    "YearMonth": "202308",
    "City/Province": "Seoul",
    "District": "All",
    "Households covered": "4,838,055",
    "Average household electricity usage (kWh)": "340",
    "Average household electricity bill (KRW)": "52,895"
  },
  {
    "YearMonth": "202508",
    "City/Province": "Seoul",
    "District": "All",
    "Households covered": "4,935,553",
    "Average household electricity usage (kWh)": "370",
    "Average household electricity bill (KRW)": "60,145"
  }
]

### 📊 When a regional electricity usage analysis is requested
When the user requests an analysis of electricity usage by region:
1. Understand the user's request.
- If the user explicitly specified conditions: use those conditions.
- If not specified: infer from the conversation context or ask the user.

2. Call "List rows present in a table" to collect the raw Seoul regional electricity usage table.
3. Pass the collected raw data, the user's request, and context to the "Time Series Data Analysis Prompt" tool. *Important!: You must enter the raw data exactly as-is.
4. Deliver the generated answer to the user.

I need an analysis of electricity usage in Dongdaemun-gu. I want to know how much it differs from Seoul overall.