Infrastructure developers are amongst the most asset-rich owners. They also generate enormous data during the project planning phase but are amongst the laggards when it comes to the use of business analytics. They are way behind industries such as retail, automotive and financial services. That is why their decision-making is too often qualitative rather than based on hard evidence. In order to embrace this technique towards a predictive system, infra companies must adopt a three-phase approach. It starts with design and data ingestion, followed by the proof of concept and then scaling. At the first phase, an exploratory analysis must be done. For this the data warehousing has to be done, as otherwise it will not be in the organized form the company would want. Only then will it be able to perform complex tasks such as correlation, natural-language processing, signal processing, outlier removal and data merge. At the next, algorithms need to be finalized and models validated. Now regression, clustering and network analysis will be possible. Permutation and holdout tests need to be performed. Before the final phase, an initial pilot has to be geared up. To close it proper scaling must be defined as the aims. The company must also ensure regular model maintenance.


Uploaded Date:23 June 2018

SKYLINE Knowledge Centre

Phone: 9971700059,9810877385
© 2017 SKYLINE. All right Reserved.