Secrets To Clinch Victory in Global Data Science Competitions

Data scientists are often perceived as crazy IT nerds who would use formulas and algorithms to even determine how many teaspoons of sugar to put in their tea! Well, we would not argue about this, as much as stereotypical this may sound, but a data scientist feels a rush when he solves a problem with calculations, analysis and logic; a rush that incomparable to anything else.

Just as an avid gamer who plays COD (Call of Duty) or CS (Counter Strike) waits for WCG (World Cyber Games). A data analyst waits for – Datahack. People who are just crazy about machine learning wait for the whole year to participate in Datahack. For them this is Olympics of Data Analysis.

To participate in Datahack, one must go through the “Smart Recruits” round. So, here are a few tricks that will help you make the best last minute strategies.

You need to get a unique perspective for building the most efficient ML Models. The competitions held within Datahack are highly fast paced ones and last only for (48 – 72 hours). Hence, you must muster thinking, acting and responding fast to find solutions to critical business problems. The winning Data Analysts are sharp on their feet (or rather minds).

Tips to win Datahack:

• You must understand that our minds never sleep. When we sleep at night it secretly works on our problems. Hence, it is a good plan to start work on a problem as early as possible because then your brain will get at least one night to work on it subconsciously.
• If you have nobody to give you a pep talk then seek inspiration from motivating videos or books. That way your mind will go beyond its usual limits. Nalin Pasricha a former Datahack winner revealed that before one hackathon he read the book “The Wright Brothers” by David McCullough which is a story of two poor brothers who manufactured bicycles. They had no funding or college education, yet they still managed to build the world’s first aeroplane. An amazing invention which scientists and universities alike could not create. He further stated that he did exceptionally well in that hackathon as his complete mindset was changed due to this inspiring tale.
• He also suggested that one should use a package or language that is new, this way your creativity will be enhanced as you would be forced to think differently.
• You must concentrate of feature engineering as it will help a lot in increasing your score.
• You should create a solid validation strategy, as without this a hackathon competition is nothing more than gambling, hence, it is important to have a good validation strategy. Know that public LB can misleading at times.
• And finally always perform ensembling or stacking, as this is the last step that will help you cover that extra last mile.
• Make sure that you choose the algorithm that is most suited for the data. Do not lose confidence on your handcrafted cross validation results.

Here were some useful tips from former winners of Datahack, now that you know these secrets it is time you put them to test and develop a habit of winning.

Good luck and may the power of data be with you…

Interested in a career in Data Analyst?

To learn more about Data Analyst with Advanced excel course – click here.
To learn more about Data Analyst with SAS Course – click here.
To learn more about Data Analyst with R Course – click here.
To learn more about Big Data Course – click here.