Weekly Statistical Forecasting model example shows how to upload customized product & location hierarchies & historical data to generate forecasts using 30 statistical algorithms.

Anaplan's Generic Weekly Statistical Forecasting model example for multiple use cases allows you to upload a customized product and location hierarchies from flat CSV files, load historical data, and generate forecasts using various statistical algorithms. The model includes 30 of these algorithms across 4 different overarching forecasting methods: Basic and Intermittent Demand, Curve Fit, Smoothing, and Seasonal Smoothing.

Not only does this model generate statistical forecasts based on historical data but it also analyzes which algorithm would best fit that data. This approach provides you with a suggestion of which method may be the most accurate to use for future periods.

Brief Intro and Overview of new statistical forecasting calculation engine model

Clear sample hierarchy and historical data, download templates to enter your data, then upload those files as CSVs to statistically forecast.
View each item's history and suggested, best-fit forecast.
Analyze the forecast and historical forecast accuracy for multiple statistical algorithms against each lowest level item.
View upper and lower bounds by standard deviation or inter-quartile range and the resulting outlier values.

Features

Product Hierarchy Management

  • Upload product data using a downloadable template and also map the uploaded data into the product hierarchy.

Customer List Management

  • Upload and manage a list of customers using a downloadable CSV template.

Historical Data Upload

  • Upload historical data for each item to be forecasted from your hierarchy also with the help of a downloadable template.

Flat List Management

  • Flexibility to set up top x Product/Customer combinations and complete CoV analysis if forecasting at multiple levels or import data as flat list otherwise.

Exception Management

  • Validate all master data elements needed by viewing potential data errors and set up default forecast settings

Outlier Classification

  • Automated Outlier Correction based on user defined parameters such as standard deviation or the Inter-Quartile Range (IQR) either for all items or for a single item.
  • Manual Outlier Correction by making manual adjustments to history.

Product Lifecycle Management

  • Slow Moving Product - identification and forecast set up by setting up zero period threshold and choose to use Croston's method for forecasting.
  • End of Life - identify trending % threshold to get End of Product suggestions and setup EOL profile to phase products out.
  • New Product Introduction - Use life profiling to generate future forecast for new items that do not have sales history.

Alternate Product History Setup

  • Set up similar product history or string history together for up to 3 products and apply to a selected product.

Override History Start Date

  • Ability to set up periods to be counted as history/forecast for each product.

Best Fit Statistical Forecasting

  • 30 different statistical algorithms overarching 4 different types of methods - Curve Fit, Smoothing, Seasonal Smoothing , Basic and Intermittent Demand Methods.
  • Ability to review forecast settings, provide manual forecast input if and where necessary and make forecast adjustments with a comprehensive view of descriptive statistics.

Forecast Algorithm Analysis

  • Forecast Accuracy analysis using MAPE , RMSE, MAD and Bias% of each of the 29 forecasting algorithms at different levels of forecast.

Descriptive Statistics

  • Ability to review the descriptive statistics for both Updated History and Adjusted History to make forecasting decisions.

Information

Size

3.34 GB

Language

English

Stats

Modules

163

Roles

3

Formulas

1751

Reports

35

Complexity

Master