ISSS608 2017-18 T1 Assign DENG CHUNLING
Contents
Objective & Methodology
In light of the serious situation that Smartpolis faces (several deaths reported!), I need to:
1. Source: Determine origin of disease outbreak
2. Spread: Find out medium of transmission
3. Control: Suggest measures to contain spread
My approach to this problem is:
Action Step | Result |
---|---|
Filter blog text for spots of flu | Exclude non-disease blogs from analysis |
Determine type of symptoms | Categorise symptoms into water, air or human |
Correlate spots with map and time | Animate disease outbreak path by timelapse |
Drill down into water-borne | Study the origin, spread and contributing factor |
Drill down into air-borne | Study the origin, spread and contributing factor |
Drill down into human-transmitted | Study the origin, spread and contributing factor |
So that I can suggest containment measures and geo-fencing for each of the transmission type.
Data Preparation
Efforts are needed to transform the "Microblog" dataset into a format that is conducive for visualization.
Variable | Treatment | Description | Screenshot |
---|---|---|---|
Location | Break into "Lat" and "Lon" respectively | SAS EG workflow to prepare Lat/Lon and save to library | |
Text | Find stemmed words from text | SAS EM workflow to extract stemmed words from library. This will help avoid fuzzy lookup e.g. I search for "ache" but "mustache" is returned | |
Text | Filter microblogs to only pre-defined symptoms | SAS EG workflow to join documents and words list. Note that these are all stemmed words so whatever forms of words are in the microblogs, even misspelled ones, will be detected. | |
Symptom | Group symptoms into 3 broad categories: related to digestive system, respiratory system or muscle. The rationale is that some people may know that they have flu, but others are just describing the symptoms. | ddd |
Origin and Spread
Transmission and Containment
Transmission Medium - Water, Air or Human Interaction?
Containment Suggestions
Link to Tableau Page
Here