Artificial Intelligence - Demand Forecasting
Demand Forecasting Using Artificial Intelligence in the latest generation of products, artificial intelligence is adding intelligence pretty much everywhere you look.
Future demand of product or a service is predicted based on the past events and prevailing trends in the present. To get accurate sales forecasts in a supply chain is certainly an important key.
If the future demand is predicted accurately it will affect other productivity factor such as planning, performance and profit of product.
In this study we had researched on effective artificial intelligence models to predict demand of customer’s product. For forecasting demand, we require good historical data. These data are required for training and validating data. Tools are needed to take advantage of it.
Artificial intelligence that allows the computer to “learn” from data. We will predict demand by implementing best artificial intelligence algorithm.
Algorithm will focus on finding the optimal structure for demand forecasting. When it comes to demand forecasting, artificial intelligence can be especially helpful in complex scenarios, allowing planners to do a much better job of forecasting difficult situations. It leverages the knowledge, experience, and skills of planners and other experts in a highly efficient and effective way across a broad range of data.
The main contribution of our work is the use of artificial intelligence models in order to predict the consumer's demand and implement this demand forecasting in a two-echelon supply chain with a game theoretic approach.
To improve product forecasts and demand plans
To allow collaboration between all departments involved in demand planning
To provide accurate demand forecast • To provide accurate cost / benefit analysis
To increase profitability Input
Unit sales price (input): Unit sales price is a competitive factor affecting the customer behaviors, especially for independent retailers. It is processed as quantitative information.
Product quality (input): This factor includes the evaluation about product quality according to customers via a 1-9 scale. It is processed as qualitative information.
Customer satisfaction level (input): This factor shows the sales and post-sales behaviors of retailers to the customers. It is processed as qualitative information.
Effect of promotions, holidays and special days (input): This factor means the percent increase of sales related to promotions or special days (such as feasts, new year’s day, etc.). It is processed as qualitative information. Output
Demand quantity (output): Demand quantity is the quantity from customers to retailers. It is processed as quantitative information. Algorithm Used
We will use Multi-Layer Perceptron (MLP) and Backpropagation algorithm which will be used to train the data
However, we will concentrate on nets with units arranged in layers
Multi-Layer Perceptron Methodology works on non-linear separable data
Following are steps of algorithm o Initialize weights at random, choose a learning rate n o Until network is trained:
For each training example (input pattern and target outputs):
➢ Present inputs for the first pattern to the input layer
➢ Sum the weighted inputs to the next layer and calculate their activations using activation function formula
➢ Present activations to the next layer, repeating (2) until the activations of the output layer are known
➢ Compare output activations to the target values for the pattern and calculate deltas for the output
➢ Propagate error backwards by using the output layer deltas to calculate the deltas for the previous layer
➢ Use these deltas to calculate those of the previous layer, repeating until the first layer is reached
➢ Calculate the weight changes for all weights and biases (treat biases as weights from a unit having an activation of 1)
➢ If training by pattern, update all the weights and biases, else repeat the cycle for all patterns, summing the changes and applying at the end of the epoch Here are five areas where we have seen Artificial Intelligence deployed specifically to improve demand forecasting
Trade promotions and media events
New product introduction (NPI)
Social listening (social media)
Extreme or complex seasonality
Trade promotions and media events Promotions, advertising are expensive, yet their impact is challenging. But with baseline demand it is difficult for the expertise to produce and predict accurate demand forecast. To solve this problem, artificial intelligence provides multitude of attributes, ranging from product and market to social activity. This technique recognizes the shared characteristics of promotional events and identifies their effect on normal sales. Multidimensional modelling that handles both qualitative and quantitative variables is particularly well suited to describe and predict the non-linear demand driven by promotional activity.
New product introduction (NPI) Here it is tough to forecast demand for a product without a sales history. With the help of artificial intelligence, you can cluster the behaviors of past launches, select the most probable performance for the new product, then “learn” common demand behaviors in the first launch period through detailed demand profiles.
Social Listening Traditional demand planning mostly depends on transactional data. But social listening can be used by the supply chain team to correlate social sentiment with demand signals. Marketing departments can know how their brand is perceived. Social channels help to enhance supply chain planning. Here we can monitor and store live tweets on specific brands.
Extreme or complex seasonality Artificial Intelligence models helps to analyze and track seasonality patterns and trends. For demand forecast system seasonality is one of the important factor to be considered.
Weather data Demand forecast depends on factors such as geographic area, products and demand lags. Artificial Intelligence can crunch that data. Artificial Intelligence can let you use weather forecasting the way you evaluate causal factors like pricing and traffic—to get the best picture of demand for a particular product during a specific time series.