By Sailesh Dhungana

Imagine having knowledge about the future. Let’s say that you know which team will win a baseball game that has not been played yet. You can place high-stake bets on it and get handsome rewards when the team wins. Knowledge about the future is a very valuable resource. Alas, the future is uncertain, and you can never be sure about what will happen next. However, you can come close. Although you will never be sure, you can guess that an event will happen in the future with a reasonable probability. This is what predictive analytics gives us. Predictive analytics helps us make educated guesses about the future.

To make guesses about the future, we need information from the past. To begin predictive analytics, we need to gather lots of data. Only after having a large number of data points where we know what the result is, we can use the prediction-making algorithms to make new predictions with new data points. For example: after having all historical records of two teams playing under various conditions, we can get the prediction of the match by passing the conditions of the next match.

The algorithms that make the prediction are written using statistics, modeling, and machine learning. Using these tools, the computer looks at historical data points and results, and makes predictions about new data points. The algorithms, however, do not always work correctly. Depending on the algorithms and the kind of analysis we are doing, there could be varying degree of errors. Predicting game outcomes might have a lot of errors, as there are many variables associated with a game.

Although there are errors and the algorithms have just begun developing, this is a very interesting topic. I have been working with Whittaker Associates to create a new predictive analytics software that identifies companies of interest. It has been a very challenging but interesting project. After all, who would not want to know the future?