Browse Prior Art Database

Method and System for Managing End to End Demand across Stores, Geographies, and Business Units by Testing Multiple Modelling Techniques

IP.com Disclosure Number: IPCOM000254250D
Publication Date: 2018-Jun-14
Document File: 5 page(s) / 190K

Publishing Venue

The IP.com Prior Art Database

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Method and System for Managing End to End Demand across Stores, Geographies, and Business Units by Testing Multiple Modelling Techniques

For businesses, demand management refers to the process of balancing supply and demand. Companies use forecasts to determine future demand and then build plans accordingly. Current sales forecasting and demand management systems do not cover cross-industry standard process for data mining (CRISP-DM) framework completely and lack end to end automation. There is a need for an end to end demand management system to predict sales, plan marketing efforts and optimize inventory, and enable users to plan demand management at a various granular level such as stores, geography, business units etc. Disclosed is a method and system managing end to end demand across stores, geographies, and business units by testing multiple modelling techniques, wherein the modeling techniques include AutoRegressive Integrated Moving Average (ARIMA), Neural Networks (NN), Linear regression, Decision trees etc. In an embodiment, the method and system comprises of exploratory data analysis (EDA) tool, modeling section and user dashboard. As exemplarily illustrated in figure 1 below, the EDA tool of the method and system runs thousands of iterations that helps a user understand the data being dealt with. Output from this section includes data summary and possible drivers of sale.

Figure 1

The method and system utilizes a storage device that contains all the data gathered, a computer or server that contains SPSS modeler and Cognos tools installed to process information, build a predictive model and develop visualization. Initially, data is read from a storage device by SPSS modeler. This data is stored in internal memory of the EDA tool until the end of modeling section. Then, the output data is moved to Cognos software using the SPSS modeler and Cognos connectivity. This information is hosted

on a server using network connectivity. The user can view the results using a computer that is connected to the network. As illustrated in Figure 1, the EDA tool provides several options to upload dataset in a predetermined fashion such as import from text file, import from database and import from csv. The EDA tool asks the user to input sales data in the required granularity i.e. hourly, daily, weekly or monthly along with the additional/optional information as illustrated in Figure 2 below.

Figure 2

Upon the user clicking EDA button, following operations are executed, not necessarily in the same order. As illustrated in Figure 1 above, univariate analysis helps to understand quantity and quality of data (check for missing values), variance, outliers, inherent trends, and seasonality. Bivariate analysis helps to check correlation between dependent and independent variables. Treatment of missing values provides several treatment options to impute missing values, including but not limited to, drop, replace with mean, replace with mode, replace with z...