The fans of famous TV series “Castle” would agree. For those who came late – In the show, arguably the hottest NYPD detective ever lived on the planet earth, Kate Becket and her team, solve cases with the help of past history records , matching the finger prints , going through the credit card statement and extensive data mining about their suspects to arrive at conclusive results. For example, Why would a bartender would suddenly have 20,000$ in his account and purchase a Maserati, given the fact – in the last 3 years, his average account balance was in the range of 2000$-2500$ or why would a Banker who always eat at fancy restaurants with his family and clients and pay from Plastic, would go to a shop joint downtown, alone and withdraw 5000$ from an ATM there. Sometimes they also predict the escape route of the suspect, by putting his trajectory into a complex predictive model which in turns gives them a short list of destinations out of 1000, where they suspect might be heading to.
Now this is TV, where the imagination can go as wild as possible, defying all the laws of physics, chemistry and of course statistics! One can argue that this is something which is not practical, possible In a real scenario, too good to be true and can be only seen in theatres (who can forget Spielberg‟s masterpiece Minority report). But this phenomenon is happening and it‟s not something new, law enforcement agencies abroad are capturing data extensively, predicting events/crimes and acting faster with the help of advanced analytics.
What, How and all of that:
The idea looks pretty simple on the surface, just like the retail giants such as Amazon Wal-Mart, use predictive analytics to find a trend in the customer‟s shopping, we can find pattern of crimes. i.e. the technique used, areas which require more attention, the time zone in which the crime rate is higher or when the crime is most likely to happen etc. In simple words, we are using analytics to analyze data, such as the times and locations of past crimes, to forecast where and when certain crimes are likely to happen in the future so our forces can stop them before they occur. The spots can be identified on the basis of mathematical models (the kind which are used in the algorithm used to predict earthquakes), and the Police forces can be prepared for any illegal activity in the area, in advance. (As the research shows that criminals tend to commit crime in proximity to the places they have commit crime. They found it as their „safe space‟ and as they haven‟t been caught in that particular area in the past, they tend to return to that are in the future).
The idea behind predictive analytics against crime is to capture as many variations possible, for ex. information related 100/911 calls. The interactions between past and forecasted information is analyzed over various time indicators such as the motive of the crime, time , weather, economic conditions of that particular time, pay day proximity, and even political factors. Every factor has its own impact on the outcome of the model and the information coming out of these models help the forces to be deployed accordingly and ultimately prevent crime.
Spiderman can’t swing on his web in India, can Predictive Analytics?
Most of the police stations in our country still follow the manual way of recording data and that involves a lot of paper work. Making some sense of all that information requires a lot of manpower with huge amount of time. But if we want a repository of all that data online (like they have in the US), to analysis it at a faster rate, where we should go? Most of us don‟t know that there is a separate agency in our country, which is recording the crime related data from last 28 years. Welcome to National Crime Records Bureau (NCRB), founded in 1986. It functions as clearing house information on crime and criminals including those operating at National and International levels. NCRB is now using GIS based analytics services to predict and curb the crime rates. It has its very own fingerprints, sketch and all other sort of database, apart from this NCRB also maintains Prison statistics. Now prison statistics can be used very useful, it can give the investigators an idea of prison inmates, who are more likely to „churn‟ after completing their time inside, on the contrary it can recognize patterns of good behaviour too, and that can help in judicial decision making process. After all it‟s all about eradicating crime and not the criminals.
In some cases even an isolated event can also be prevented, like what happened in Muzaffarnagar riots, the live feed from social networking websites can be integrated in the model and it can inform, the law enforcement agencies about the unusual behaviour of the crowd in any particular area or about any particular topics, i.e. if they are sharing or commenting on a picture or video or it is being shared too many times with suspicious tags in a very short interval of time.
Wait! What’s the catch?
Besides getting accurate data, can only be used in cases where there is a pattern. Now not every crime that is being committed out there follows a pattern but yes, a significant amount of all sorts of crimes, are being done in such a way that a pattern can be made. Analytics can be especially useful in cases like auto theft and burglary, where patterns can be detected. According to Mark Cleverly, who heads the IBM unit for predictive crime analytics, “You can build a model that factors in attributes like the time of year, whether it is hot and humid or cold and snowy, if it is a payday when people are carrying a lot of cash,”. Now to be clear it doesn‟t mean to forecast accurately that a crime will occur at a fixed time and place, that‟s practically impossible (as of now). But it can surely give accurate insights about the time window and a certain parameter of the city, which is more likely to have a higher crime rate.
In the end:
Talking about numbers, In the city of Memphis only, serious crimes fell to 30 percent and while other violent crimes went down to 15 percent after implementing predictive analytics. Now this was in 2006. Since then the data being captured has reached to a humongous amount and now the new, more accurate techniques and tools are available for the predictions. If implemented properly, we are looking at 45-50% reduction in crime, all over the world. Now this will take time and lots of patience, but for what little can be seen from this point is that, the idea of a Utopian Nation is very much achievable and we don‟t need any Avengers or Justice League for that, we will write our own destiny.
If you loved this story, do join our Telegram Community.
Also, you can write for us and be one of the 500+ experts who have contributed stories at AIM. Share your nominations here.
Besides being a daydreamer, a self proclaimed know-it-all jackass and terrible writer, Ritesh M Srivastava completed his engineering from Amity University and did his Post Graduate Program in Business Analytics from Praxis Business School. He has over 4 years of work experience, handling and surviving different technologies, currently working as Senior Associate in BI&AS (Business Intelligence and Analytics Services) division of 3i-infotech India.