Know The Reason Why Data Analytics Is Heavy Over Data Engineering?

By Jyoti Nigania | Dec 12, 2018 | 1953 Views

While several firms have commenced data analytics initiatives, solely some are victorious. Studies have shown that 70% of data analytics programs fail to understand their full potential and over 80% of the digital transformation initiatives fail. Whereas there are several reasons that have an effect on victorious readying of data analytics, one elementary reason is lack of fine quality data. 
However, several business enterprises understand this and invest considerable time and energy in data cleansing and remediation, technically referred to as data engineering. It's calculable that approximately 60 to 70% of the hassle in data analytics is on data engineering. As long as data quality is a vital demand for analytics, there are five key reasons why data analytics is significant on data engineering:
1. Different systems and technical mechanisms to integrate data:
Business systems are designed and enforced for a purpose; primarily for recording business transactions. The mechanisms for data capture in Business systems like ERP is discrete data whereas within the SCADA/IoT systems it's for continuous/time-series data. This implies that these business systems store various data varieties caused by the speed, volume, and selection dimensions within the data. Hence the technology (including the information itself) to capture data is varied and complicated. And after you are attempting to integrate data from these various systems from totally different vendors, the data model varies leading to data integration challenges.
2. Totally different time frames of data capture
The timeframes for data bodily process throughout data capture varies. as an example, in ERP systems the info bodily process is usually batch/discrete/manual, whereas in SCADA/IoT systems, the info bodily process is typically automatic and time period. as an example, once the merchandise delivery to the client is completed, the cargo details are usually captured in the time period by the hand-held devices. however the invoicing can't be directly processed as invoices are issued from the ERP systems to the client. This creates a delay in Delivery-Invoicing compliance reportage.
3. Totally different user value-propositions
In business, the identical data is made and consumed by totally different stakeholders (inside the company) in numerous ways that as their value-propositions vary. as an example, seller payment terms for Finance could be a value object, whereas for acquisition the identical data component could be a risk component (as longer payment terms typically end in longer deliveries).
4. Totally different business processes
The same data component is totally different because of variations in business processes supported geographies, laws, laws, market conditions, etc. as an example, the data component in the North American nation is subject to data privacy laws, whereas data component in most developing countries is mostly not a part of the info privacy laws. So, obtaining client shopping for habit report supported age for a developing market is far "easier" than obtaining the identical report in North American nation. 
5. Totally different aggregations driven by structure structures
One data component is viewed otherwise supported variations in granularities or aggregations driven by structure structures. as an example, the VP of acquisition would possibly like a pay report supported item classes (an aggregation of things procured), whereas the acquisition manager wants the pay report supported individual things procured. That aggregation would possibly vary supported the item sort, provider sort, delivery location, etc.
Good analytics depends on sensible quality data. So, if you're embarking on the analytics journey by viewing technologies, tools, and hiring data scientists, pause for a moment. Challenge your assumption and raise one basic question is a diversity of my business operations poignant honest quality data for analytics? If the solution is affirmative, prepare for a protracted and a fancy data engineering effort.

Source: HOB