File Upload Documentation

Overview

This document outlines the two main file upload processes in the application:

1. Sales File Upload
2. Item List File Upload

Each process serves different purposes and has specific implications for the system's data and model training.

Sales File Upload

Upload Methods

The Sales file can be uploaded through the "Upload Sales" page using two distinct methods:

1. Upload and Replace

   • Initializes the system with completely new data
   • Removes all previous sales data
   • Triggers new model training from scratch with the uploaded data

2. Upload and Combine

   • Attempts to merge new data with existing data
   • Maintains duplicate prevention mechanisms
   • May encounter errors if there are inconsistencies between:
     • Data formatting
     • Column names
     • Data structure

Column Mapping

After uploading a sales file, a mapping dialog automatically appears to ensure correct data interpretation:

• The system displays required columns and allows you to map them to your file's column names
• This accommodates different naming conventions in source files (e.g., "OrderDate" vs "OrderDateTime")
• Previously used mappings are remembered for convenience in future uploads
• You can manually adjust any incorrect mappings before processing
• Processing will not begin until all required columns are properly mapped

Training Status Monitoring

After file upload, the system provides real-time training status information:

1. Status Indicators

   • Training status badge shows current state (e.g., "TRAINING")
   • Colored dot indicator on the navigation menu shows active training

2. Timestamp Information

   • Last status update time
   • Last file upload date and time

This allows you to monitor model training progress after uploading new sales data without needing to refresh the page manually.

Processed Information

The system processes the following fields from the Sales file:

Field	Description	Requirements
OrderDateTime	Date of the order	Also accepts RequirementDate
Quantity	Main prediction target	Required
MeasureUnit	Unit of measurement	Items with "pieces" are primary focus for predictions
ItemCode	Grouping identifier	Required
ItemClassification	Used for classification predictions	Required
ItemDescription	Display name in UI	Required
Standard	Inclusion flag for model predictions	Required

Any cell that has value of 1 in standard will be flagged for item to include in model training. However, if any item that is flagged for training but less then 10 sales history, its filtered and excluded from items to include in model training. This is done because of the lack of data to make a reliable predictive model.

Impact on Item Configurations

When the Sales file contains a Standard field:

• Updates data in "Sales Options" page under "Item Configurations"
• Items with Standard = 1 are automatically flagged as "Always Included" for model training
• For items present in sales but missing in Item Configuration:
  • Creates default configuration with:
    • Lead Time: 28 days
    • Target Cumulative: 100%

Item List File Upload

File Structure

Required columns:

• ItemCode
• LeadTime
• TargetCumulative

Optional columns:

• ItemClass (This information will appear on All export excel file Supply Planning sheet.)

Behavior

• Completely replaces existing item configuration data
• Purpose: Updates LeadTime and TargetCumulative values
• Prevents retention of outdated default configurations from previous sales uploads
• New sales uploads will still create default configurations for missing items

The Item Options can later on removed if desired with the button. It is also possible to manually select items to include in training process.

Important Notes

• The system maintains data integrity by preventing duplicates during combined uploads
• Default configurations are created dynamically when needed
• Item List uploads help maintain accurate LeadTime and TargetCumulative values. It is not required to updated before each sales data upload, but it is recommended if sales data includes an item of interest that is not included in Item List to get accurate LeadTime and TargetCumulative values.
• The two upload processes work together to maintain system data accuracy