Building the Forecast Dataset

Every strong forecast starts with well-structured source data.

What's in the Dataset

The file contains 1,500 rows of order-level transactions spanning three full years (2022–2024), across 5 regions, 5 product categories, 4 sales channels, and 3 customer segments. Each row represents one order and records 21 fields:

Why This Structure Matters for Forecasting

A well-designed dataset gives forecasting models three things they depend on:

• Granularity — row-level detail lets you aggregate any way you need: by month, by region, by channel, or any combination;
• Consistency — every row uses the same 21 columns, so SUMIF, SUMIFS, and PivotTable formulas work reliably across all 1,500 records without exceptions or workarounds;
• Time anchoring — having Order_Date, Year, Quarter, Month, and Month_Number already present means you can build time-series analyses immediately, without extra wrangling.

Task: Explore the Dataset

Goal: Get fully oriented in the workbook before touching any formulas.

1. Open Business_Performance_Data.xlsx and navigate to the main data sheet.
2. Press Ctrl+End to jump to the last used cell — confirm you land on row 1501 (header + 1,500 data rows) and column U.
3. Click cell A1, then press Ctrl+Right to hop across all 21 column headers. Read each name and match it to the table above.
4. In a blank cell, type =COUNTA(A:A)-1 to verify the row count programmatically. It should return 1500.
5. Select the full dataset and press Ctrl+Shift+L to enable filters. Use the filter dropdowns to explore the unique values in Region, Product_Category, Sales_Channel, and Customer_Segment.
6. In a blank cell, enter =SUM(O:O) and note the total revenue figure — you'll use this as a cross-check throughout the course.