Abstract

A lightweight, interactive Python desktop app that predicts future monthly sales using linear regression and recent average trends. Designed for analysts, it requires no complex setup: just copy-paste your Excel data, click, and visualize 12-month forecasts in a dynamic chart. The tool makes forecasting accessible, interpretable, and exportable to Excel in seconds—ideal for business planning & performance tracking.

Methodology

• Data Input

  • Users copy and paste monthly sales periods (e.g., Jan-2023, Feb-2023, etc.) and corresponding sales values from Excel.
  • The number of past months to include in the average calculation is user-defined (default is 6).

• Data Processing

  • The date strings are converted to datetime format using pandas.to_datetime().
  • Dates are transformed into numeric format using matplotlib.dates.date2num() to fit the regression model.

• Forecasting

  • A linear regression model is trained using scikit-learn's LinearRegression.
  • Features (X): Numeric representation of past dates.
  • Target (y): Monthly sales figures.
  • Forecasted values for the next 12 months are predicted using the model.
  • A simple average of the most recent n months is also calculated to serve as a baseline comparator.

• Visualization

  • Historical sales, regression line, forecast points, and average line are plotted using matplotlib.
  • Sales values are annotated on the graph to enhance interpretability.
  • The graph is embedded into a pop-up window within the Tkinter GUI.

• Output

  • A result table displays past sales, forecasted sales, and the average benchmark in a horizontally scrollable table.
  • The output can be exported to Excel for further analysis.

Results

• The application has been tested on sample datasets and demonstrates:
• Clear Forecast Trends: Users can observe the direction and slope of future projections.
• Visual Validation: Overlays of historical and forecasted data support intuitive validation.
• Export Capability: The Excel output allows integration with other business reports or dashboards.

Potential Applications

• Monthly revenue forecasting
• Demand planning and procurement
• Retail inventory predictions
• Finance/Operations KPI tracking
• Budget projection comparisons

Limitations

• Only linear models: Non-linear trends (seasonality, abrupt change) may not be well captured.