Predictions
As long as humans have been around, we have always been trying to understand out environment around us. As we are starting to get a grasp on a few aspects around us, we realized that certain patters tend to be followed by a certain result, which can lead to a prediction. These patterns can be simple enough to decipher instantly, or complex enough to require a large amount of code to parse through it.
Humans have been trying to predict the future not only to feed our curiosity, but also to better our self preservation. Modern predictions have saved countless lives from natural disasters, disease and much more. A collaborative community has formed around predicting things for the common good, and my hometown of Chicago has a pretty healthy community of hackers for the common good. One thing I went to on Tuesday nights downtown was Chi Hack Night, where people came together to work on open source projects that benefited the city and people of Chicago. Many of these happened to be data driven predictive models that I would love to highlight here.
Chicago Clear Water: http://chicago.github.io/clear-water/. This project predicts the water quality at major beaches in Chicago based off of past records and weather data. The problem is that the traditional testing methods that are usually used to determine if a beach is safe for swimming does not return the results in a timely manner. Time matters in the middle of a Chicago summer, when it seems like the whole city is at the beach, so this model returns surprising accurate results before the water actually becomes dangerous.
The approach is that they regularly test 5 beaches that account for over 50% of the contaminated water. The rest of the beaches can be separated into clusters, and one beach per cluster can be tested to determine the levels for the entire cluster. It allows for more concentrated testing of waters, which saves time and allows for more tests to be done in areas that are probably contaminated.
West Nile Virus Prediction Model https://github.com/Chicago/west-nile-virus-predictions. This model predicts the likelihood that mosquitos in a certain area will test positive for West Nile Virus if the area tested positive for West Nile Virus last week. It can do this with an amazing 80% accuracy. It used the presence of West Nile Virus previously, weather and location of the areas to predict if the area will test positive for West Nile Virus.
The most important part of these projects are the data. Fortunately enough, Chicago, like many major cities, has a data portal at https://data.cityofchicago.org/. Here, you can find data on everything from crime to economics to transportation. This data can then be cleaned, formatted and used to create either predictive models, apps or websites to help the public in everyday life. The rise of open data has been a huge catalyst in the rise of citizen hackers, and it has raised the overall wellbeing of many areas as a direct result!
