Roshann.my

"Hello, World!" was my 1st computer program, 10 months ago as I began my journey in data science. Learning from online & offline resources, I grew exponentially as a professional to face digitalisation.

For your benefit, I’ve written 10 key-takeaways on insights I gained from being a data scientist:

1. Never too old to code, never too smart to code

Coding (a.k.a. programming) is prerequisite to be a data scientist. The good news is, learning to code is easier now than it was 1 nor 10 years ago. There’s plenty of programming languages to choose from, plus the newer ones are simpler yet feature the powerful capabilities of older complex languages. Python (most popular language) would take you just a couple of hours to grasp the basics. Heck you don’t even need a computer - I learnt Python on the SoloLearn App on my iPhone during my commutes to work. If you can afford classes to learn programming, by all means go ahead. Only ignorance is bliss, which will result in you blaming everything but yourself when something goes wrong (hint: while/for/if loop). With the advent of polyglot programming, you only need to know 1 programming language as there are tools that allow different languages e.g. C++ to speak to Python code.

2. Problem? Diagnose it for options to solve 

Debugging is not exactly fun, especially when you’ve spent days writing a program… only for it to fail unexpectedly during the demo to your client. In this results-oriented world, everyone wants to see the final product,  the elegant & beautiful website (a.k.a. dashboard). When you’re involved in the iterations, you realise that 80% of website glitches are related to the database which relates to the models running the show & data coming in. The next time you have a problem with your computer or smartphone, try identifying what’s the root cause of the problem. This will involve a lot of trials, reading & distinguishing advice against crappy solutions. You can decide to get someone to solve it, but you’ll miss out the learning required to keep up when the next problem comes by.

3. Not paying for a product? You’re the product!

In my opinion, there are currently 3 business models being used extensively online:

1. Paid – Self-explanatory, you pay for a service, receive customer service & data privacy is prioritised. (e.g. Microsoft Office 365, Spotify (premium), dJay Pro)


2. Open Source– No payment required, you can download codes for Firefox & completely make it “yours” i.e. changing the code to suit your needs. Open Source programs are normally funded by voluntary donations. If you’re keen, you may contribute to the development by learning the code or participating in feedback discussions to share any ideas which you have. (e.g. Mozilla Firefox, Wikipedia, SoundCloud, OpenDNS)


3. Free – You’ve seen it. “FREE – no payment”, sign-in now with Facebook & Google (we won’t post on these platforms). Let’s face the ugly truth: building an app requires long hours, smart people & a lot of money. To gain user tracking, in the front end it is “FREE”. On the back end (i.e. Database), your moves (X-Analytics – how you’re using your trackpad & mouse) are being used to train data science models. This training data eventually builds up into AI tailored to advertising needs – to make money from you. Pay close attention to the ads you’re getting, see if you need to start over. Thread with caution when using a free app, before it starts using you. (e.g. Google, Facebook, Instagram, Twitter, Snapchat, Spotify with Ads)

4. Call bluffs when faced with statistical lies

The 2 graphs below are using the same exact data – yet at a glance, they seem to tell a different story. I enjoy looking at graphs in reputable sources like The Economist, Wall Street Journal & Time Magazine, which strive to give an unbiased view of the story covered (refer RIGHT example below).

In addition, be precautious when dealing with “average”. Average could be:

Mean – the sum of the values divided by the number of values
Median – the "middle" value separating the higher half from the lower half
Mode – the values that appear the most in the dataset

Learning statistics will allow open your mind and eyes to shine when faced with bias. Credibility is the cornerstone in today’s media.

5. Too good to be true? It is too good to be true!

Desperate times call for desperate measures… by desperate people usually. The most common example being the Nigerian Prince scam… you know the one, you probably had one in your mailbox. If someone promises you $1 million to provide your bank details, you will not disclose your details. On the same note, do not blindly share your salary information for “career assessment purposes”. Know what you’re worth & ask for what you’re worth. If a recruitment consultant says “$4k is too high, what’s your last drawn salary?” You say “OK, bye bye”. Lowballing may not be the best approach to securing a job.

Fact: There are always jobs available, provided you’re able to create value. If you can bring in $5k worth of value to the company, you may negotiate for $4k perhaps. But if you can’t carry your salary, you won’t last long… so don’t worry too much about it. Digital recruitment is increasingly becoming a numbers game as both prospective employers & recruitment consultants have access to the same recruitment platform i.e. LinkedIn. Trust yourself and decide.

6. Break bottlenecks to score the Goal

The pace needed to keep up in today’s world is getting faster. So fast, that it makes you wonder if you can keep up? You give your all, nothing’s left on your plate & it’s only 3pm on a Wednesday. “Am I not doing enough?” you ponder. Bottlenecks prevent liquid from coming out all at once, resulting in a slow release of the liquid not matter which angle you turn the bottle. In a workforce with 4 generations at the moment, Baby boomers, Gen-X, Gen-Y & Gen-Z… there will be bottlenecks that prevent you from flowing at your pace. When you have 1 week to analyse a dataset for practical insights, don’t spend 4.5 days gathering data alone. If you spend more than 1 day due to a bottleneck identified, escalate the matter immediately & not on the last day of the week. Your time in this world is finite, spend your minutes & be productive. The goal today seems to be increasingly about value creation & value in itself is highly subjective. Pursue your own initiatives when you can & you’ll be the top scorer winning the game for your team!

7. Give a human touch with machine learning

A computer does 2 things way better than us humans can possibly imagine:

Counting & Remembering

Q1: How many times did you smile this year?

A1: If you have a camera recording your every smile, you’ll have the answer in a second.

Q2: Which hour of the day did you smile the most?

A2: By sorting your data, you’ll have the answer in a heartbeat.

Q3: Why did I smile the most at 6pm every day this year?

A3: The computer will not be able to answer this.

Human touch: 6pm is when I get off work each day, so that’s probably it! I’m happy when I get to go home!

As a data scientist, data is a tool which you can use to uncover insights. But computers are not the best at coming up with the right questions, to get the right insights supported with the right data. Judgement, wisdom, experience & a little pinch of luck may help discover the next biggest answer in service of our human race.

8. We’re up all night to get lucky

The agile way of working involves sprinting. Employees at Unicorn’s like Uber did not start out working from 9am to 5pm to revolutionise the way we see transportation today. They sprinted 80 hour weeks to develop their minimum viable product (MVP) to go market as fast as they could, before anyone else came along. They gambled 80 hours of no pay, only employee stock options, in dreams that Uber will have an Initial Public Offering (IPO). Failure was not an option to these brave souls who gambled their careers & retired young & wealthy enough – to enjoy the finer things in life. Work hard, work smart.

9. Dance with music, not noisemakers

There has never been a time in history where we have access to so much beautiful music. On the flipside, some still prefer noise otherwise known as the hive mentality. As human beings, we have beautiful minds that allow us to think, judge & feel if something makes sense & also if there’s a sense of style. There is an ocean of “fake news” on social media & blogs. In data science, we believe in the theory called “garbage in, garbage out”. LinkedIn, Facebook, Instagram & Twitter are a graveyard of fake profiles which are sold to boost “credibility” of those who can afford it. Do not embarrass yourself by supporting noisemakers, unless you are attracted to noise. Machine learning will also prioritise your learning experience based on the content you interact with. So, choose life… before selling your soul to a model driving your “success”.

10. Passion + Perseverance + Persistence = Perfect iterations

Finally, I’d like to share on my personal beliefs. The digital world will make our world a better place, we can all learn from each other at a pace that no historical figure ever imagined possible. Our smartphones make us like cyborgs, able to perform “miracles” at a touch of our fingertips. In our career as professionals, passion is still of upmost importance. If you're not passionate about working & like to produce music instead, it’s as easy as downloading an app on your computer & voila - you have the same tools that Calvin Harris uses to make music. Perseverance is next, as learning takes self-initiative and restraint from unproductive activities. Nevertheless, learning is still easier now than it ever has been in history. Persistence is about standing up to let your voice be heard. Have principles & most importantly, believe in yourself - that's self-confidence. With passion, perseverance & persistence, you’ll be able to aim for perfection through continuous iterations of you being a better you each day, in & out.

Let’s work & pray together for a digital renaissance.