Episode 2: Finance Data extract - Unveiling the secrets

When it comes to applying data science to finance data, one of the crucial initial steps is data extraction. Extracting the right data sets the foundation for insightful analysis and informed decision-making. In this article, we'll explore the importance of data extraction in the context of financial data and its role in enabling effective modeling.

Unlocking Finance Data:

Finance data is typically housed within finance systems, with transactional data residing in finance ERPs and specific subledger data stored in corresponding ledgers. For example, fixed assets data can be found in the fixed asset module. However, for modeling purposes at the corporate level, starting with aggregated data is often sufficient. Consolidation systems often provide translated data into the parent currency, making it an ideal source for extraction.

The Manual Extract Advantage:

Starting with manual data extracts, rather than diving into automation, is a recommended approach. This is because modeling, including machine learning, involves a trial-and-error process. As you progress, you may discover new data points and fine-tune your models. Automating prematurely without this iterative process can result in wasted time and effort. Therefore, manual extracts offer flexibility and adaptability during the initial stages.

The Power of CSV Format:

Choosing the right format for extracted data is essential, and CSV (Comma-Separated Values) format often proves advantageous. CSV files are easy to handle and transform, simplifying the data preparation phase. Additionally, ML models tend to work seamlessly with CSV data, making it a preferred choice for compatibility and ease of understanding.

A Use Case: Predicting Revenue for a Services Company:

Let's consider a practical use case where a services company aims to predict revenue, with a significant focus on the impact of marketing activities (ignoring other factors for this example). To begin, the revenue data needs to be extracted. While balances can be obtained from the General Ledger (GL), it may not provide detailed-level data. To overcome this limitation, extracting more granular data, such as per customer data, can yield richer insights. Managing the data size while increasing granularity is key to striking the right balance.

Obtaining Marketing Data:

To incorporate marketing data into the analysis, the first step is to acquire the necessary data. Marketing expenses can be extracted from the accounting system, especially if it includes a channel breakdown. However, if such breakdowns are unavailable, extracting detailed marketing data from a dedicated Marketing ERP, including channel-specific information and timeframes, becomes crucial.

Data Cleanup: An Essential Exercise:

Before feeding the extracted data into subsequent systems or models, a crucial step is data cleanup. In subsequent episodes, we will delve into the intricacies of financial data cleanup, discussing best practices and techniques to ensure data integrity and reliability.