(photo by Roy Chang)
One of the cornerstones of machine learning and predictive analytics is the availability of relevant data. Typically, 'big data' refers to vast amounts of data that challenge the existing software tools in terms of being manageable. As artificial intelligence (AI) tools are developed for project management the availability of data is crucial to the reduction of error rates and consequently obtaining successful results.
What kinds of data might be relevant for project management? Arguably the most important documents are the lessons learned from similar projects and the risk register with an assessment of the effectiveness of the risk response strategy. There are two main factors in acquiring this data:
1) How much data can be acquired? (i.e., How large is the sample size?)
2) Will the documents be available?
Public data such as economic data (inflation, unemployment rates) or publicly advertised data (the price of gasoline or utilities) is more easily obtained so this article deals with the issues for accessing non-public data. The acquisition of non-public data will therefore include data is that is available within an organization or possibly through a network of partners.
AI typically requires a large sample size in order to be more accurate. https://en.wikipedia.org/wiki/Law_of_large_numbers
This favours large organizations and organizations that complete many projects since all of the internal data can then be made available to future projects. Smaller organizations or those that have projects on an infrequent basis are at a disadvantage. For non-profit organizations such as associations there may be a requirement to create partnerships for sharing project data. For smaller businesses in a competitive environment, this may become a problem. First, they do not have vast amounts of project experience and second they are likely to avoid sharing project results due to security or business privacy concerns. A small corporation that completes 5 or 6 projects a year will probably not have sufficient data for successful machine learning. Therefore, they need to search for partners in a similar industry or performing similar projects to take advantage of cross pollination for valid samples. Access to this data will be challenging if the industry consists of competitors because data about projects can reveal numerous internal attributes about an organization that they may want to remain private.
This scenario means that large organizations are better positioned to provide the required datasets and take advantage of AI development than smaller organizations. The result is that large organizations have further competitive advantage over smaller businesses and this has the potential to change the business ecosystem. The challenge for AI developers in project management will be to find a way for all organizations to benefit from technological advances.