miércoles, 15 de junio de 2016

How intelligent forecast for SAP® Business One HANA will help Fashion companies to run better.

The reality that is now gradually being accepted both by those who work in the industry and those who research forecasting is that the demand for fashion products cannot be forecast. Instead companies need to recognize that fashion markets are complex open systems that frequently demonstrate high levels of ‘chaos’. In such conditions, managerial efforts may be better expended on devising strategies and structures that enable products to be created, manufactured and delivered on the basis of ‘realtime’ demand

The difficulty in predicting demand has led companies to focus on the improvement of the supply chain. This is one of the factors in the success of brands such as H&M and Zara, which have the shortest market lead times.

Ok then…., does it make sense to continue to study demand forecasting?

The purpose of this blog is to analyses the fashion product characteristics and also the demand planning variables to support the use of the intelligent forecast algorithms provided in HANA.

Analysis for fashion products

These are the main characteristics typically exhibited by products in the fashion industry.

1      Short lifecycles: the product is often ephemeral, designed to capture the mood of the moment; consequently, the period in which it will be saleable is likely to be very short and seasonal, measured in months or even weeks.
2.      Short selling season: today’s fashion market place is highly competitive and the constant need to ‘refresh’ product ranges means that there is an inevitable move by many retailers to extend the number of ‘seasons’, i.e., year. The implications of this trend for supply chain management are clearly profound.
3.      Long replenishment lead times. Regarding the features of the demand:
4.      High impulse purchasing: many buying decisions by consumers for these products are made at the point of purchase.
5.      High volatility: demand for these products is rarely stable or linear. It may be influenced by the vagaries of weather, films, or even by pop stars and footballers. There are numerous sources of uncertainty in a fashion supply pipeline, starting with demand through to the reliability on the part of suppliers and shippers, etc.
6.      Low predictability: because of the volatility of demand it is extremely difficult to forecast with any accuracy even total demand within a period, let alone weekbyweek or itembyitem demand.
7.      Tremendous product variety: demand is now more fragmented and the consumer more discerning about quality and choice.
8.      Large variance in demand and high number of stock keeping units: as a result, the volume of sales per SKU is very low and demand for SKUs within the same product line can vary significantly.

Analysis of demand

A lot of literature regarding demand planning features affecting forecast such as smoothness, intermittence, slow-moving and so on. Basically these are not or partially applicable in the fashion businesses. Only Lumpiness which is defined as:  the feature of tending to have periods of very low or zero demand and then spikes of demand. A lumpy demand is variable, sporadic and nervous.

Data Aggregation concept

The fashion industry mainly needs forecasts at two levels of data aggregation:

      The “family level” composed of items of same category (T-Shirts, trousers, . . . )
which enables companies to plan and to schedule purchase, production and supply at mid term. For this aggregation level, historical data usually exist.

The “SKU level” which is required to replenish and to allocate inventory in stores at a shorter horizon. At this level, references (SKU) are ephemeral since they are created for only one season. Thus, historical data are not available, even if many items more or less similar have usually been sold in previous seasons.

Forecasting Methods without Historical Data

Most of fashion items are sold during only one season. Companies have to estimate the sales without any historical data: the forecasting system should be then designed for new product sales forecasting. New product forecasting is one of the most difficult forecasting problem. Indeed, standard forecasting methods are not suitable. In this context, a two-step methodology seems emerged:

1. To cluster and to classify new products to forecast their sales profile (mid-term forecast).

2. To adapt and to readjust this profile according to the first weeks of sales (short term forecast).

If no historical data exists for the considered item, but similar products have already been sold in previous seasons. Indeed, new products usually replace old ones with almost the same style and/or functionality (i.e. T-shirt, pull over, .), it is thus possible to use historical data of similar products to estimate the sales profile of the new products.
Thus, to forecast the sales profiles of new products such as garments with clustering and classification techniques, descriptive attributes (price, life span, salesperiod, style, . . . ) of historical and new products should be taken into account. The aim is to model the relationship between historical data, i.e. between sales and descriptive criteria of related items, and then to use these relationships to forecast future sales from descriptive criteria of new items. These relationships are often complex and non-linear. For this kind of problem, machine learning methods have demonstrated their efficiency for building simple and interpretable pattern classification models

About the SAP® Business One Intelligent Forecast

SAP Business One, particularly SAP Business One version for SAP HANA, is packed full of tools for storing and interpreting data in the Predictive Analysis Library

The Intelligent Forecast is a statistical forecast tool with built-in models incorporating trends and seasonal factors. It includes 2 forecasting methods. SAP Business One automatically selects the best algorithm:

TESM stands for Triple Exponential Smoothing. TESM is used to handle time series data containing seasonal components. It works by incorporating a stationary component, trends, and seasonal factors. Both the trend and seasonal factors can be additive or multiplicative in nature.


LRDTSA stands for Linear Regression with Damped Trend and Seasonal Adjust. It is chosen for forecasting when times series data presents a trend. A damped smoothing parameter is used to smooth forecasted values and prevent over-casting. This method also detects seasonality in your data in order to adjust your forecasting results.

The following study will focus how to implement the fashion aggregation and clustering concept to the SAP® Business One HANA Intelligent Forecast. 

Copyright © Argentis Consulting / www.sapapparel.com/