Dear Team,
Remember , " If you don't make yourself obsolete , then
someone else will " ?
Even as we sit comfortably in our cocoon , some headhunting firm
somewhere in India , is planning to employ BIG
DATA , to render us obsolete !
hcp
--------------------------------------------------------------------------------------------
India can become world's data science service provider
( source : MINT / 01 July 2016 )
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
In the last two decades, we
have seen a huge impact of information technology (IT) on businesses. Today,
almost all business transactions and processes, internal and external, happen
on a computer or mobile and by the use of a network, primarily the Internet.
The penetration of smart phones
has extended this automation to the last mile i.e. to the consumers.
Large network-based information
systems facilitate various business processes such as sales order, financial
transactions, human resource management,
customer service management and so on—the Aadhaar project being a splendid
example of IT reaching the last mile.
The IT revolution has made businesses much more efficient by
allowing transactions to be fast, error-free and trackable.
It has not necessarily made
them ‘intelligent’ but it has paved the way for mining
intelligence by creating a wealth of digitized data. For instance,
product manufacturers know what product is sold on a given day, at what time
and at which outlet. A lot of transactions happen naturally on the web, through
e-commerce.
In the coming decade, i.e. the decade of data science, we will analyse
these data on a large scale to identify trends,
find anomalies and most importantly, predict future trends.
But let us pause here to
understand what data science is.
Suppose you have the
transcripts of the pitch of sales people
in a company including the ones that have led to a sale. You wish to predict which sales pitches are good and what makes them good.
Traditionally, one will
probably use a sales coach to understand this.
However, the data science way is remarkably different—it uses
unstructured data, i.e. transcripts, and derives features such as ,
* the length of the call,
* counts of courteous expressions like “may I”, “please”;
* counts of words about product value—“automatic”, “fast” and so
on.
Using these features and
statistics, an algorithm builds a model to predict
the success of the pitch.
For instance, the algorithm
may discover that ,
* pitches with a good number of positive words,
* good number of product value words and
* with a moderate call length leads to success.
A totally new sales pitch can then be checked against such
conditions to predict if it is good or not.
------------------------------------------------------------------------------------------------------------------------
My comments :
# A " Resume
" is a jobseeker's " Sales Pitch
" ( just as a " Job Advt " is an employer's " Sales Pitch
" )
# By plotting frequency diagram of " total no of words " contained in ( say )
, 10,000 resumes , we can
find the " Mean " ( Read my handwritten note " The Beauty Index " written sometimes in 2003
)
# Positive
Words = Keywords ( belonging to a
given " Function " ) contained in a given resume ( after removing
duplicates ) , whose "
weightage " we take into account , to compute " Raw Score " of any given resume
The higher the Raw Score , the
better the " Sales Pitch "
# Using RESUME RATER , even "
Job Advertisements " ( or " Job Descriptions " provided by a
Client against a
mandate ) can be rated
to find its " Raw Score " . Then compare this with the " Raw
Scores " of " Sales Pitches "
( ie " Resumes
" ) of each executive being " considered / shortlisted " for
that assignment
# Over the past 5/6 years , we
started keeping meticulous records in File Finder
, of all Mandates and
against each mandate ,
* shortlisted
resumes
* interviewed
candidates
* appointed
candidates
* joined
candidates .....etc
If we were to subject
this vast data to BIG DATA analytics ,
then I believe , we would get unprecedented "
Insights " , which would enable us to " predict " ( within given probabilities )
, which of the candidates ( being
considered against any
given new mandate ) has how much " Chance (
ie Probability ) of being accepted at
each of the
above-mentioned " Stages " of
headhunting process
I am likely to get some
help from IBM
Watson Analytics Team , to undertake a BIG
DATA ANALYTICS of over
2 million job advts that we have compiled in CustomizeResume.com
database , over the past 7 years
When this project gets
over ( as a new page on CustomizeResume web site ) , I will discuss with IBM ,
the
possibility of what I
have suggested above
If any one as any
suggestion to make , please do so
hcp /
02 July 2016
--------------------------------------------------------------------------------------------------------------------------------
How is this revolutionary?
We can find out whether a new sales person is ready to be put on the
job or needs further training based on his/her pitch quality.
This model can be linked to
the sales process, where it provides personalized feedback for improvement to
each salesperson immediately after a sales call.
Eventually, one day, the
algorithm will replace the salesperson!
Better algorithms, more processing power and the availability of
large data sets have enabled highly accurate models. However, this process of
data science itself is hardly automated.
Initially, a lot of work is
required to convert unstructured data in to useful models and integration of
these models into the information systems. The person responsible for
all of this is the formidable data scientist.
Can India provide an army of data scientists to the world, like
the IT engineers that it provided in the last decade? Can this become the
engine of our economic growth in the coming decade? It is not only a huge
opportunity, but a challenge, too.
First, we need trained manpower. Data scientists need to be adept
in programming skills, database skills and basic statistics. Today, we hardly
have people who understand both computer science and statistics. Our
undergraduate education needs to take up data science courses in a big way. A
year ago, we took a bold step of introducing data science for classes V to
VIII. These kids successfully built their own friends’ predictor (www.datasciencekids.org). India
should aspire to be a leader in providing data science education early on.
Second, we need to be among the leaders in research and innovation
in machine learning. Though India became an IT services powerhouse without any
great technology innovation, this will not hold for data science services, for
innovation and disruption have gathered rapid pace in recent years.
We need to use the latest technology in our services and also be
the innovators. Data science service offering will not only be people driven
but also product driven. Unfortunately, there is a huge gap here. Our analysis
(ml-india.org) showed that Tsinghua
University in China alone produces more machine learning papers in top
conferences than all universities in India put together. Similarly, companies
in the US have taken a lead in machine learning innovation.
Lastly and most importantly, we need entrepreneurs who can go
across the globe and solicit data science business for their companies in
India—the new Murthys and Premjis. Let us get ourselves ready to become the
world’s data science services powerhouse.
----------------------------------------------------------------------------------------------------------------------------
Varun Aggarwal is
co-founder and chief technology officer of Aspiring
Minds, a company that uses data science to enable meritocracy in the labour market.
No comments:
Post a Comment