We want to create a repository of apps / APIs that provide forecasts and predictions for a broad class of questions that concern all human beings. We would also like to create our own predictive API - so if you are interested in this project, contact us at [email protected], we might have a gig for you! Here's an example of what we did in the past. We also designed apps for stock market predictions (see our data science book, page 195-208) and copyright infringement detection.
Basically, we see two types of apps that are based on data science:
- Predictive apps that require little user input (predicting weather, Earthquakes. diseases etc. where input is Zipcode, maybe a few extra fields such as gender or age)
- Processing data from clients via an API to return scores for each transaction: the API fetches the (possibly large) data via HTTP (data is stored on client's server) then process data, for instance score transactions - and store results on a server, ready to be accessed by client via machine-to-machine communications. Parameters are provided through the API to tell whether the data is used for learning (machine learning, training set) or prediction, and the API behaves differently based on these input parameters (model refinment, or generating predictions, depending on the inputparameters).
Here we are interested in #1, though we are also in the process of designing a generic predictor based on HDT and Jackknife regression, with source code available at no cost for everyone (open intellectual property).
Predictive apps currently available
This is a starting point. Please let us know (in the comment section below) if there are other apps you'd like to share, or provide a link. Here are a few ones to begin with:
- Predicting the value of your home
- Predicting your risk of stroke / dying / cancer etc in the next 5 years based on a few indicators (smoking status etc.)
- Predicting your risk of divorce
- Predicting your number of Twitter followers
- Predicting Earthquakes, weather, tsunamis, solar flares, wars
- Predicting your chance of losing your job, or your future job performance and salary
- Predicting your chance of tax audit, being arrested, going bankrupt, becoming millionaire, being sued, or car accident
- Predicting arrival time (driving or flying from A to B)
- Predicting your risk of catching measles (based on location, vaccination status, age, health conditions etc.)
- Predicting sales or traffic arising from various marketing campaigns
- Predictive extreme floogs per location
- Predicting prices of food crops, hotel rooms, metals, energy sources
- Predicting the chance of a meteoroid impact
I would imagined many of these apps would be based on data collected via crowdourscing, using decision trees for the predictive engine. I would think that some of these apps would be offered for free, with monetization via advertising.
Finally, designing this type of apps can be a lucrative jobs for data scientists, consisting in selling or licensing refined data (that you collected) or predictions.
Good to read