Limitation

Study shows: biased police-recorded data

from To predict and serve? by Kristian Lum, William Isaac

• If police focuses on certain races, police records over-represent the certain racial groups. What this means is that crimes that occur in the same locations are more likely to appear in the database
• "Bias in police records can also be attributed to levels of community trust in police, and the desired amount of local policing – both of which can be expected to vary according to geographic location and the demographic make-up of communities. " from To predict and serve? by Kristian Lum, William Isaac

• Tay - Microsoft's automated Chatbot @TayandYou
  • Trained Tay with misogynistic and offensive tweets, then became part of the data corpus used to train Tay
  • In a day, Tay's Twitter account was put on hold because of Tay was posting similarly offensive tweets.

• Google Flu Trends
  • "a near real-time service that purported to infer the intensity and location of influenza outbreaks by applying machine learning models to search volume data." from To predict and serve? by Kristian Lum, William Isaac
  • The models completely missed the influenza A–H1N1 occurred in 2009 pandemic and overpredicting the flu from 2011 to 2014.

Hypothesis

• Our hypotheses using machine learning algorithms were to predict where a next crime will happen and the type of the next crime at a certain time period but they did not work
• "The data is collected as a by-product of police activity, predictions made on the basis of patterns learned from this data do not pertain to future instances of crime on the whole." from To predict and serve? by Kristian Lum, William Isaac