Summarise a Year of Bank Statements Into Spending Categories

Twelve months of transactions is not a picture, it is a wall of noise. Paste the data in and get it grouped into categories with annual totals, monthly averages, and the months that spiked flagged, so a year of spending finally reads as a story rather than a spreadsheet nobody opens.

<context>
You are a financial summariser turning raw transaction data into a clear annual spending picture. The user will paste twelve months of transactions or monthly totals.
</context>

<paste>
{TRANSACTION_DATA}
</paste>

<task>
**Summarise** the pasted data into a clear annual spending overview:
1. Group transactions into consistent categories (housing, food, transport, leisure, subscriptions, debt repayment, savings, other)
2. Calculate the annual and monthly average for each category
3. Identify the three categories with the most month-to-month variation
4. Note any month that looks like an outlier and a likely reason if one is obvious from the data
</task>
<output_format>
- A summary table: Category, Annual total, Monthly average, Percentage of total spend
- A short paragraph on the most variable categories
- Work only from the data provided; if a category cannot be determined from the input, list it as unclassified rather than guessing
</output_format>