As I walked in the seminar room of IBA Karachi`s exquisitely-built CED building, I didn’t expect such a huge crowd of highly professional looking individuals and students gathered to witness the talk on data science, a topic usually of interest to geeks alone.
I took my seat among the audience and as the crowd settled down, a bearded man wearing spectacles stepped up towards the podium. This man was, Mr. Zeeshan-ul-Hassan Usmani, a genius of social entrepreneurship, a renowned Pakistani data scientist, writer, entrepreneur, and scholar in the field of computer science.
Mr. Usmani, broke the ice, telling the audience that his computer had crashed this morning, so he had to produce a whole new presentation, therefore, we would have to bear with a few problems here and there. Nevertheless, he progressed on and straight ahead, dived into the reason we were all there: to hear him talk about Data Science and PredictifyMe.
Who is Zeeshan-ul-Hassan Usmani?
For those who are not familiar with his name, he is one of the handful data scientists and one of the front-runners of entrepreneurship produced by Pakistan. Hailing from a small town Sukkur and the youngest of 14 siblings, he was the only one to take full advantage of his education. We covered a story on his journey a while ago, and we have also had the pleasure of having him as a contributor to TechJuice.
A Fulbright scholar, holding an MS and PhD degree from the Florida Institute of Technology (FIT) and the founder of PredictifyMe, a predictive analytics company that uses big data to solve social as well as economic problems, he has been featured on numerous publications around the world, including but not limited to: WSJ, CNN, TechinAsia and FloridaToday! His interview for the CNN-Money is really a must-watch, where he leaves the interviewer flabbergasted by the information he was able to tell about her, simply by using publicly-available data. He even went as far as to foretell the exact number of views, likes, shares and twitter mentions his very own interview was going to get!.
You can check out the interview here.
The Potential of Data Science
“Privacy doesn’t exist anymore!”- Mr Usmani remarked during the start of his talk highlighting the importance of data science in today’s world. If someone wants to know something about you, he will. To prove that, he conducted an experiment with Halloween Candy and as a result, has put forward a thought-provoking revelation that he can successfully predict some really private info based on the candy you give out to kids as trick-or-treat. What private info? Well for starters, he can tell who you are going to vote for, the age of your wife and even the number of daughters you have.
These kind of demonstrations show how “Big data” can be used in to solve problems related to almost every practical field of life. Mr. Usmani went on to explain that here are half a million publicly available data sets for every American and 37,000 distinct values can be found about every American by simply utilizing this data. He stated that as of now, an average person(in the U.S) produces around 1 T.B of data per year but in a few years, this will become tenfolds to 10 TB of data per annum! This shows a big potential for data scientists as it will be creating a lot of data that needs to be sorted through to find the right things and as a result, the demand for data scientists is going to sky-rocket.
His small startup, PredictifyMe, which consists of only 53 employees right now, recently raised over $1.25 Million in funding to delve into the fields of data science.
Suicide Bombings in Pakistan
Starting off as a doctoral work for his PhD degree on big data, his work evolved and led him to create his own company, Predictify Me with the the philosophy of “code for life”. He believes that “every line of code we write should have an impact on someone’s life around the world”. This is in line with his mission of prevention of suicide bombings which he has been pursuing tirelessly for the past 10 years! Talking about his motivations to work in the area, he said,
“There is a suicide bombing somewhere in the world every 33 hours and a suicide bombing in Pakistan every 6th day. There is an IED attack in the world every 11 hours…I’ve asked myself what as a computer scientist I can do to solve this problem. BlastSim came out of this thought in 8 years.”
Unlike U.S’s way of using classified information, Mr.Usmani took advantage of the vast number of freely-available data from public records: Hospitals, Police, Media, etc. in his work. Due to the lax laws in Pakistan over data, he also knows confidential details like where the bomber was standing, what shrapnel he used and where he exploded.
“[We] have the largest data set on earth when it comes to suicide bombings”, he said.
Through this data, he has managed to come up with his own injury model, a Usmani-Kirk model that can be used to detect the exact amount of explosives, casualties, injuries, etc. with upto 90% accuracy using the software “Blast-Sim” and can be used to predict the same. The software is one of a kind, having the option to fill in almost every single detail you can think of like the height, weight of bomber, temperature of environment, details about the bomb(weight/amount of explosive, type of bomb), nearby cars(and even the amount of petrol in them), trees, number of people around, and a whole lot of other options. There is even an option to evaluate data in 3D!
“If, for example, a suicide bombing happened and all you learn is that 13 people were killed and another 12 were injured, my software can regenerate the whole scene for you—just like in a CSI: Miami episode.” – he claims.
Adding to its complexity, medical records of people can be used to predict if the impact of the bomb from where they are standing will kill them or not. Mr Usmani highlighted the massive implications that can be derived by data analysis, covering but not restricted to:
- Minimizing damage by suicide attack: In Pakistan, according to Mr. Usmani, only 25 kilograms of explosives can be carried by the bomber with good concealment, an amount only effective within 22 metres of range. This makes them recommend to building authorities to make the security gate at least 30 metres away from the building.
- Triage of Patients: Using the software, accurate predictions can be made about the injuries of the patients to ensure fast treatment.
- Forensic Analysis: He claims to determine who conducted the attack with an accuracy of 93% just by examining the medical/legal reports of the patients, as every terrorist organization has a different blast signature.
- Survivors: In case of a lot of post-explosion debris, this software can determine the chance of remaining life among the bomb-victims, and allow the emergency teams to dig in the right place where there is a high-chance of survivors.
Narrating an incidence where he demonstrated the software, back in December of 2009, when a tragic bomb blast on a Shi’ite procession on the day of Ashura was immediately termed as a suicide bomb attack; he was the first one to claim that be not! Sure enough, a few days later, it turned out to be a remote-controlled bomb attack. Javed Akbar Riaz, superintendent of police for the area of Karachi where the blast occurred, commended Mr. Usmani’s help in determining the nature of attack.
Some Statistics from Pakistan
His company analyzed a total of 129-threat cities in Pakistan and here are some eye-opening statistics about suicide-bombing in Pakistan,
- In Pakistan, one IED attack occurs every 67 hours.
- March-September is the most-dangerous period in terms of suicide bombings, the tendency reaching maximum in September. There have been 33 suicide bomb attacks in just the past 3 years in the month of September.
- Peshawar has the highest number of attacks.
- Very few suicide bomb attacks have ever been recorded in the Islamic month of Zil-Haj.
- Tehrik-e Taliban Pakistan(TTP) doesn’t conduct suicide bombings in Balochistan.
- September 22 is the most dangerous date! Almost every year, there is a suicide bomb on this date.
- Sectarian attacks have decreased in frequency in the last few years, but there intensity has increased drastically.
- Thursday is the favorite day for suicide bombers.
Zeeshan Usmani is also the founder of PakistanBodyCount.org, a website dedicated to providing accurate statistics about suicide attacks as well as drone attacks.
Predicting A Suicide Bombing Before it has Happened
With their whole treasure-trove of data, Zeeshan Usmani boasts a 72% accuracy in predicting a suicide bombing before it even happens. Although he can’t pinpoint the exact location yet, his team is working on it. As of now, he can predict the attack days before the attack happens using a software called Soothsayer, developed at PredictifyME. Out of a total 19 predictions made by his company, 15 proved to be true, which is a big achievement and is miles ahead of other publicly-available softwares.
Soothsayer makes the prediction by analyzing a total of 200 indicators including weather, sporting events, major holidays, attacks in nearby countries, visits by international dignitaries and the emergence and popularity of a blasphemous video on YouTube or Facebook. Numerous factors are taken into account to make a pattern out of the factors while predicting an attack.
“There might be a heat wave, extreme cold weather, or there might be something else. For example, TNT is often used in Pakistan in the mining industry. Now, summer months like June and July are very, very hot, so the mining industry is not active in those months. So if you examine blast patterns, you do not find TNT being used as an explosive in those months because TNT is not readily available due to the seasonal halt of the mining industry. That’s a pattern right there. The second thing is we have more bombs in the month of March and September. These are the months for plowing the fields, so the terrorists use the calcium ammonium nitrate used in agriculture—these are the so-called “fertilizer bombs.” (The Brown Journal of World Affairs)
In a sense, Zeeshan Usmani, along his team of data scientists, is like a modern Sherlock Holmes for suicide bomb attacks, re-enacting the whole environment with bits and pieces of data, and then extracting the essential information. Only in this case, he doesn’t only extract the information about the culprit, but can also predict the next moves in advance and then recommend the safety measures to protect against such attacks. Genius!
On the topic of how many schools in Pakistan are using PredictifyMe’s softwares, he said they have started in Pakistan with one school a while ago, and are looking to increase that number to 1,000. They also plan to launch this program in another countries including Nigeria, Sudan and Lebanon.
Discussing his future plans with our team, Mr. Usmani seemed hopeful to look into worldwide school shootings after successful implementation of this program, starting from Pakistan and the US. Meanwhile, his company PredictifyMe is working on a product called HourGlass, a data analytical tool to generate its own “artificial big data”.
He ended his talk by advising all the attendees to read as much as they can because that’s the only way to progress in the competitive field of today. You can access Mr.Usmani’s own books via this link.