Artificial Neural Network
Artificial neural network models (ANNs) is a relatively new approach for addressing critical analytical tasks that has received wide attention in many academic and research fields. The idea of ANN is loosely based on the biological neurons. Neurons are the most basic elements of the human brain which provides us with the abilities to remember, think,and apply previous experiences to our every action. Each of these neurons can connectwith up to multitude of other neurons. The power of brain arises from the numbersof these basic components and the multiple connections between them. Basically, abiological neuron receives inputs from various sources, combines them in some waywith respect to the stimulus, and then gives the output or the final result.
How it works
An Artificial Neural Network (ANN) is an information processing system which works similar to the human brain. It uses the human brain’s pattern while solving problems through learning and then stores that knowledge through interconnections. It is made up of a large number of highly interconnected processing elements which process information. The neural network uses processing units like neurons of the human brain to build connection strengths between the units to store the learned knowledge. The learnedknowledge then allows for the neural network to make generalizations of the outputs based on certain inputs.
Three Layer Architecture or Approach
One of the most common neural network architectures is a so-called three layerarchitecture. The first layer is called the input layer and is the only layer exposed toexternal signals. The input layer transmits signals to the nodes in the next layer, whichis called a hidden layer. The hidden layer extracts relevant features or patterns from thereceived signals. Those features or patterns that are considered important are then directedto the output layer, the final layer of the network. When compared with other statistical methods, neural networks deal better with nonlinearities in the data, large numbers of categorical variables, and complex relationships regarding inputs and provided superior prediction rates.
Neural networks have been applied in various areas of business. In finance, neuralnetworks have helped improve the field of research by revisiting traditional problemssuch as bankruptcy, financial risk management, pricing and hedging derivative securities,forecasting future currency exchange rates, and trading in emerging capital markets.
Neural networks can also help in understanding employee behavior, successful new ventures, organizational structure, and voluntary employee turnover. Managers can use the networks to predict turnover behavior. Management can help save losses in profits, jobs, and productivity by utilizing neural networks as a resource. Neural networks have been useful to help understand inventory control, advertising, new product development, and consumer behavior. In consumer decision making, neural networks have been used to predict consumer shopping behavior with great accuracy. A neural network is based on learning the complex relationships between variables such as attributes and consumer choice and therefore provides a better prediction of the decision made by the consumer.
In consumer behavior, neural networks can help researchers understand the decision process based on different attributes of the product or service and the different environmental and personal influences on the consumer such as opinion leadership, demographics, etc. In retailing, neural networks can provide forecasting information to better assist with inventories, merchandising, and negotiations with manufactures, especially with the help of scanner data. By using neural networks, a store manager can order the correct amount of product, saving the store inventory costs and building a loyal customer base through reliability. In building internet marketing strategies, neural networks can assist consumers in searching information on a firm’s website, predict future purchases made based on present purchases, and classify customers based on their behavior patterns. Companies can even customize an entire customer experience through a company website by allowing the customer to opt in, track purchases and preferences in a customer database, and allowing the neural network to predict shopping behavior. The information provided could give the company an opportunity to anticipate items the customer might want and help build loyalty to the store and online shopping. In sales management, neural networks can help profile the behavior of sales employees. Neural networks can help management understand such behavior. Sales employees can provide data that can profile their customer base to forecast customer order amounts, frequency of customer orders and their overall choice behavior for new products, trials, etc. In marketing management, neural networks can assist in multiple criteria decision making, new product teams, and pricing.
Marketing managers can use these networks to look at marketing strategies used in the past and to forecast new marketing strategies or new sales as well as pricing of new products. Managers can use neural networks to find an optimal organizational structure for new product teams and new product development decisions to ensure the success of new products. Also, neural networks can be utilized to predict the risk of joint ventures and alliances between firms. Technology, talent, knowledge, and capital can be inputs into the network to determine a successful venture. Furthermore, advertising campaigns can be chosen based on criteria chosen by management regardless of the criteria imposed on the project. These advertisements can then be analyzed to predict the effect of the advertisement on the consumer in pre-testing the market and post-testing marketing research.
Neural networks have emerged in the marketing research literature as an alternative tool to analyze data. These networks can help researchers and practitioners understand complex consumer decision-making behavior. There is a tremendous opportunity for neural networks to be used in many areas of future business and research because of its intelligent brain-like structure and its strong prediction capabilities.