Transaction classification

Transaction classification forms the foundation of any reliable analysis of bank transaction data. Data extracted from a bank’s API in raw format that is difficult to use or interpret, so categorising each transaction into recognisable groups such as Groceries or Utilities enables fast extraction of value from this data.

The most important metric in transaction classification is accuracy. Analysis built on a foundation of inaccurate data will be similarly inaccurate, which is why transaction classification accuracy is the central feature of Planky’s product.

Planky employs a suite of machine learning models and data processing algorithms to achieve the highest standards in classification accuracy. Our models are trained on millions of rows of data and categorise each new transaction into 100+ categories.

Since outputs of classification models are statistical, misclassified transactions resulting from variance in the data is inevitable, but users can expect to see the industry benchmark of 90% accuracy achieved or exceeded within Planky’s outputs.

To access classified transaction data via the dashboard, select the Reports page from the main Planky dashboard panel as shown below. This will display a list of users who have shared data (these are denoted by a ready button).

	Clicking the button for the user of your choice will take you to their report page.

To access transaction classification data, simply scroll to the Categories of Income or Categories of Expenses sections:

The transaction classification display contains the following components:

• 1. Transaction categories. Transactions are grouped by category within dropdown bars labelled by that category. In certain instances, categories are bundled together under a broader subheading (for instance, in the above, Council Tax, Mortgage/Rent, etc. are bundled under the grouping Bills). This is done simply for ease of navigation.
• 2. Transaction data. Clicking on a category at the lowest level of granularity reveals the individual transactions themselves.
• 3. The total income or expenditure, in the designated currency, for that category.
• 4. The percentage contribution by that category to the total income or expenditure across the whole account/user.
• 5. Transaction date
• 6. Transaction description
• 7. Income / Expenditure flag
• 8. Transaction amount