-
Build infrastructure for user uploads/downloads
-
CSV
- AWS S3
-
Output from Google Form
- MongoDB
-
Geojson
- AWS S3
-
Json
- MongoDB
- CartoDB dynamic updates?
- Jason
- make the data from the original sources
- store all datasets on S3 bucket
-
Make a MongoDB that holds geojsons. Takes user input (eg PCode) and inserts a sub-dict into the object's json dict.
- Callout
- Callout
- Myanmar "Table"
- Townships "Table"
- Villages "Table"
-
Accomplishments
- Completed Actions 1
- Completed Actions 2
- Completed Actions 3
- I don't know
-
Action 1
- Action 1a
- Action 1b
- Action 2
- Action 3
- Action 4
- Action 5
- Action 6
- drag files & drop here
-
Discover segments/clusters of townships in the census data
-
affluence factors
- cash
- banks
- credit cards
- cell phones
- electric power
- livelihoods/economics
- Anthony
- Who are the customers?
- We need to condense all of the census data features into a more concise set of features (PCA).
- Visualize and define clusters in a 2 or 3 dimensional state (kmeans, GMM, DBSCAN)
-
Accomplishments
- Completed Actions 1
- Completed Actions 2
- Completed Actions 3
-
Action 1
- Action 1a
- Action 1b
- Action 2
- Action 3
- Aug 20
- Action 4
- Action 5
- Action 6
- project sample document.pdf
-
Predict Ability to Pay
- Have a clear metric that enable decision makers to compare townships/villages (visually)
- Anthony
- PCA(census data matrix) => 1dimensional "affluence index"
- Random Forest/Linear Regression => Feature Importance
-
Accomplishments
- Completed Actions 1
- Completed Actions 2
- Completed Actions 3
- Next Month
-
scale all the features
- Action 1a
- Action 1b
- do PCA
- Interpret PCA vectors and feature scores
- Action 4
- Action 5
- Action 6
- drag files & drop here
-
Front End / UX
- Collect
- Describe
- Discover
- Predict
- Advise
- Cameron