Assignment #6
What happens in the following stages of predictive modeling: collect data, clean data, identify patterns, make predictions?
Collecting Data
IoT collects the most data
Increased specificity
Intimate Data
Assumptions are made based upon collective data
Real-time sharing of data
Ship / Weather data
Connecting all data to create highly complex algorithms
Clean Data
Cleaning to create a highly specific algorithm
A lot of types of unstructured data, cleaning involves structuring
More data needs interpretive systems
Identifying Patterns
Distill into something that can translate across data points
Make Predictions
Predictions can only be as strong as your data
Weak data can’t be manipulated to be stronger, additional data points need to be added
To what degree are accurate, real-time predictive analytics possible?
With the increasing usage of IoT devices, real-time data is constantly being crunched. The article mentions Waze in particular, where users can contribute real-time traffic accidents, police, etc, and the app can change a user's route based upon this information. On a larger scale, much of transportation, especially by sea relies on real-time predictive analytics. A company with a large fleet of ships that transport goods, for example, needs to be able to anticipate the weather, delays, and other information relevant to efficient, safe and fast travel from destination to destination. Given that this has become the norm, I can only imagine a world in which the majority of large-scale future events are predicted.
How do votes work in PageRank?
PageRank seems to work through an algorithm, as words and search terms crunch through the algorithm a sort of relevancy output is spit out which determines how a page ranks. The "votes" that a page gets are also partially determined by the number of "links" to a page. As explained in the article, if a highly voted page links out to another website, that site is also transferred some of the "votes" of the origin page.
To what degree does captured user data influence search results?
The article discusses that the propensity for users to click away from a page may impact ranking. By using Chrome, for example, Google is able to store the results from their browser users and reference them later. Though this may be helpful in the name of efficiency, the ability to manipulate the way we find references based upon how users interact with pages could lead to a burying of resources that are for whatever reason deemed irrelevant.
