Garcia, Jeremiah R.

Business Analytics

Business Intelligence is the applications that transform data into meaningful information which helps
business make better decision.

Brief History

The term business intelligence came into use around the 1950s, and it grew out of technology called
decision support. Over the past few decades, business intelligence systems have grown more powerful
and comprehensive, mainly due to increased data collection on their customers and on their own
internal processes. This is due to the rise in things like smartphones, wearable devices, the internet, the
general computer usage, etc. Also, companies can store data, cheaper and in greater quantities than
ever before.

Companies now have access to smartphone metadata, internet usage records, social media activity, etc.
Business intelligence platform sift through this data to find patterns and trends among purchases or
their own internal manufacturing processes. By 2018, the Business Intelligence Market is expected to be
worth $ 20.1 Billion.

There are three main forms of data, (1) structured (2) semi structured and (3) unstructured.
Unstructured data is information that can’t be easily read by computers. It’s difficult to organize in
traditional databases, because it can’t be stored in rows or columns. This is the most form of data in the
internet. A good rule of thumb is that 80% of all data produced is unstructured. For instance, when the
business collects information about Facebook usage, any kind of messages or any kind of comments on
the wall are all unstructured data

Structured data resides in a fixed form and its labeled, such as name collection forms on websites. It’s
easy for computers to read and query such information, because the data is already standardized.

How do companies manage and store all data?

Company data isn’t always in one location. It’s usually found across: CRM programs, Marketing
automation systems, Social media platforms, etc. The first stem in BI is taking inventory of the data your
company process.

Data warehouses are used to consolidate disparate data in a central location. Usonf a process known as
exact, transform, and load (ETL), warehouses standardize data across systems, which allows it to be
queried.

Data marts are essentially smaller, more focused warehouses. Instead of aggregating data across a
company, a data mert might store the information of just a single department. Data marts limit the
complexity of databases, and are cheaper to implement than full warehouses.

How does information get to a central location?

The method for standardizing and centralizing data is known as extract, transform, and load (ETL).
 1. Extract. Raw data is extracted from a source program (such as CRM or ERP software). This is
often the step where unstructured data (such as notes, or author information) is tagged with
metadata to make it easier to find.
2. Transform. During this step, the data is normalized. In order to properly analyze data, it must be
in the same format – think apples to apples.
3. Load. Data is transferred into the central warehouse or data mert. This process can occur every
week, day, hour, or every minute. The more often this is done, the more up to date analytic
reports will be.

What is Hadoop?

Hadoop is an infrastructure for storing and processing large sets of data across multiple servers. Instead
of centralizes files, Hadoop uses a cluster system that allows files to be stored on multiple servers.
Hadoop can be complex to implement and run, and isn’t well suited for ad hoc queries. Hadoop is best
suited for companies that produce massive volumes of data, such as Facebook, or eBay.

Map Reduce is the processing arm of Hadoop. It allows data to be queried and processed on the server
where it resides, instead of transporting the data across the network to be analyzed on the computer.
This can save huge amounts of network bandwidth and resources.

Analyzing Big Data

Data analysis is the reason companies invest in BI. The insights from analytics reports influence company
direction, product lineups, and even hiring decisions. Here are some of the key terms and concepts.

Data mining is the analysis of large sets of data in order to find patterns and correlations. It can be used
to: group sets of data, find outliers and draw connections. It could be argued that all the “intelligence” in
business intelligence results from data mining.

Text analytics software combs through unstructured textual data to find patterns. This is particularly
useful for analyzing the sentiment of social media posts, or online customer feedback.

Business Analytics. By analyzing and drawing connections between data, companies can: predict future
trends, gain competitive advantages and reveal unknown inefficiencies. There are three main forms of
business analytics, (1) descriptive, (2) predictive and (3) decision. Descriptive analytics analyze past data
and identify trends and relationships. Raw data can be grouped into easily digestible pieces, such as the
number of unique page-views, or the sales numbers for a specific department. Predictive analytics
searches for a correlation between a single unit or factor, and the features that pertains to it. The goal is
often to find the same correlation across different data sets, which would allow companies to infer
future patterns from past trends. This is an increasingly popular form of analytics. Decision analytics
helps companies analyze future industries and market spaces. Decision analytics looks at a company’s
internal data, then analyzes external conditions (such as manufacturing trends, or predicted supply
shortages) to recommend the best course of action for a company.

Business Intelligence Reporting

Dashboards are the interfaces that represent specific analyses. They’re particularly helpful for those who
don’t wish to interact with software through a command line interface. Market research shows that
more companies than ever before are implementing BI programs. The majority view it as a business
opportunity. However, getting access to clean, high-quality data remains difficult for some companies.

Current Trends

1. In-memory Processing

In-memory systems utilize RAM memory instead of hard drives to execute queries. This vastly increases
application performance. As a solid state memory price continues to drop, in-memory processing will
likely became the standard.

2. Usability and Visualization

Business intelligence software is making its way out of the IT department. Visualization, dashboards and
intuitive UX design allow programs to be used across departments. Companies that implement user-
friendly software throughout their workplace stand to benefit from the insights produced from such
collaboration.

Business Intelligence in Action (O2 Ireland Case Study)

O2 Ireland is a arge cell-phone carrier, and Telefonica Europe subsidiary.C Customers were buying their
prepaid SIM cards for their phones, then leaving the country. O2 Ireland didn’t want to spend marketing
resources on these customers, but didn’t have a way to segment from their marketing efforts. O2
Ireland contracted Teradata to centralize their information, and then contracted Cognos to analyze the
data. Using their new BI system, they were able to target just the customers with long-term prospects.

Business intelligence can help companies make better decisions by showing present and historical data
within their business context. Analysts can leverage BI to provide performance and competitor
benchmarks to make the organization run smoother and more efficiently. Analysts can also more easily
spot market trends to increase sales or revenue. Used effectively, the right data can help with anything
from compliance to hiring efforts.

A few ways that business intelligence can help companies make smarter, data-driven decisions: (1)
Identify ways to increase profit, (2) Analyze customer behavior, (3) Compare data with competitors, (4)
Track performance, (5) Optimize operations, (6) Predict success, (7) Spot market trends, and (7) Discover
issues or problems.