Page 27 - BCICAI-Connect-Dec-2023

Article

Talking to the Data

CA Nishith Seth

Data is something which has no meaning/ sense Once the data is properly cleansed, now the
to its user. It’s a raw form of anything whether in process of analytics starts, which shall involve use
structured or unstructured format, that can be of various commands, functions, and algorithms to
processed further to make it useful/ meaningful. achieve the answer of WHY.
Every data talk to its user, we need to listen to Standardise the basic analytics/reporting and
it. It is up to the skills and capability of the data graduate from daily working to auto mode
user, how you massage, analyze and interpret it. reporting. This forms the basis for advanced
Most of the time, it is found, it tells us something, analytics, including Machine Learning, Predictive
which the user is not aware of. To understand and Artificial Intelligence Analytics.
data and to take it to the next level of Artificial
Intelligence, we need to talk and understand what Machine Learning provides computers with
data is telling us. TALK TO DATA, IT IS WAITING FOR the ability to learn – without being overtly
YOUR AUDIENCE. programmed, meaning they can teach themselves
to grow and change when exposed to new data.
Data Analytics is a systematic journey from the Machine learning uses analytics from historical
basics to intelligence. There is an excitement data to detect patterns in new data and adjust
among the users/ consumers of the data to reach programme actions accordingly. The purpose of
to the pinnacle of the analytics but they miss to machine learning is to discover patterns in your
put the basics in place. Let us understand how to data and then make predictions based on often
start the journey and talk to the data. First, you complex findings to answer business questions,
should be clear about WHAT you want and WHY detect and analyse trends and help solve problems.
you want it. Background and Objectives of the Machine learning is effectively a method of data
analytics should be clear and documented by the analysis that works by automating the process of
user. It is much easier to write it on paper than to building data models. Machine learning examines
start hitting the ALT key over EXCEL. Once, you have small or large amounts of data possibly from
your background and objectives in place, identify many different sources with statistical algorithms
where the data is and are available in what format. such as clustering/ profiling; regression and
Data could be structured or unstructured.
classification. The objective is to discover patterns
Another challenge faced by the user is getting the and then make predictions based on those often
right and complete data on a continuous basis. complex patterns to answer business questions
Such challenges are normally resolved with the and solve problems.
support and commitment from the Management. Clustering/ Profiling – Is the task of separating
It has been seen that once the user gets the data, a set of un-labelled objects into groups such
there is a hurry to start the analytics. On the that those in one group are more similar to each
contrary, users should focus on data cleaning and other than they are to objects in other groups.
massaging, the process not only involves removing An example of a clustering problem is identifying
unwanted characters, but also requires making groups of people with similar buying patterns.
sure data is reliable for analytics.

| 27