PROJECT: The use of intelligent systems for planning and scheduling product development projects.
Introduction
Today, information technologies have become one of the most important keys of a company’s development. In fact, new product development (NPD) is the crucial process in maintaining a company’s competitive position among all competitors. NPD is a risky activity as market competition and product technology advancement are growing intensely. In reality, more than half of the 20% of successful cases failed to return expected investment cost and turned out spending more cost and time to complete.
The main reason they fail is because of the sequential processes and due to the difficulties in constructing reasonable development schedules and resource distribution plans. Therefore, these reasons indicate a need for intelligent systems that uses a database of past product development projects to estimate the duration of project phases and improving the quality of planning and scheduling.
The goal of this paper is to demonstrate the use of intelligent systems to identify the relationships in an ERP(Enterprise resource planning) database data attributes (e.g. duration of delivery, number of team members) and the duration of a project to improve product development.
Problem Description
The scheduling and planning phases are important to create good product. Since the real world data is so complex and inconsistent, traditional linear system may suffer from significant tendency in data mining. This lead to poor schedules and plans quality, which later would effect the cost of development. Therefore, intelligent systems became an important position to consider for modelling complex data mining problems.
The importance of this intelligence system is quite strong for companies that have their own products, such as tech industries, such as Apple and Samsung, or fashion brand industries, such as Gucci and Prada, that would constantly create new products. Or even private business that may produce their own product. The competitiveness level of these types of industries is strong. Companies that have better NPD are found to be produce better quality products. For tech companies, in order to be ahead of the competitors, they indeed depend on the quality and gap of time of producing new products.
Many real applications and software that provide services like planning and scheduling production cycle could be found across the Internet. There are different pros and cons for each of them. A few examples of them are Asana and Monday.com .
Data sources relevant to the problem and how data is collected and stored.
A huge amount of data has been regularly generated in many part of economy, including business processes in companies. These data sets are not only enormous but also often unstructured and complex. Analyse this data with manual methods is slow, expensive, and prone to errors. Hence, data mining techniques are needed to automate the process.
Data source
There are two types of data mining tasks, which are descriptive and predictive. The descriptive techniques simply provide the summary of the data while predictive techniques study and review the current data to make predictions about the behaviour of new data sets.
The table below compares the steps of the most used methodologies for developing data mining and knowledge discovery projects.
ERP databases often contains thousands of attributes and a huge amount of data that are irrelevant to the present mining task. The most basic rule of data analysis is simply data reduction. Hence, variable selection has to be taken carefully to allow improvements on model predictive ability, and later reduce modelling time and unwanted noises.
Variable selection aims to construct a model that could predicts well and explains the relationships among the data as simple as possible. There are a lot of dimension reduction methods. Some of the examples are Mallows Cp, Bayesian Information Criterion (BIC) and Akaike (AIC), Factor Analysis, Projection Pursuit and Principal Component Analysis.
Use of false application of data mining methods will only draw more worthless and incorrect patterns. The method of estimating project duration includes two data mining techniques, which are neural networks and fuzzy sets.
The figure below shows the estimation of project duration with the use of fuzzy neural system.
Method of project duration estimation by intelligent systems
Planning and scheduling of project activities are the crucial factors for a new product development to meet the project requirement. This step uses the eliminators of the required resources, duration of overall project process, their sequences, and such on. Apparently, the quality of these eliminators may depend on the type of new product along with the techniques used for estimating. Products that has their very own potential, for example new technology that are new to the world and will create their own market, are hard to obtain a high-quality forecast. However, tech companies would further consider the use of modification of current product to create new products.
The following method is dedicated to the company-typed enterprises that use an ERP system to run their business processes like the development of new product. The stages of product development always depend on the features and characteristics of the product and the company in which it is designed to.
However, some common phases can be useful, for example, concept development, planning, detail design, testing and refinement, system-level design, and production ramp-up. These phases can also be considered in the context of pilot, concept initiation, prototype, program approval, and launch. To acquire a project schedule, the duration of project phase and specification of resource availability is required. These parameters can be estimated with the use of an ERP database or defined by experts.
The first approach is suitable for projects that have a unique form, for example, innovations and construction projects. In turn, if a company is developing something out of modifications of previous products, then it is very likely to recognize the patterns from the ERP database and to use them to improve the estimation quality of project planning and scheduling.
Data visualization
The input variables of the example phase in the product development project are shown below.
These variables derive from concern departments and an enterprise’s internal database (ERP system) such as production, purchasing, materials management, project management, and, research and development (R&D) . The development of a product prototype
includes many processes that require, for instance, storage of materials, the purchase of materials from the suppliers and usage of materials in production.
The figure below shows the stages of the proposed method.
(Analytic lifecycle)
Conclusion
New product development is the crucial process in maintaining a company’s competitive position in this technology advanced era. Product development is, however, a complicated and difficult process. It could be expensive and yet risky for the stability of the company. Nevertheless, bad scheduling and planning of project development is the main reason a product fail to be developed. The use of intelligent system is needed to generate better schedule plans. With the help of various data mining techniques and fuzzy neural system, product development schedule is improved and can be visualized. Efficiency is the key in successful product development.