Using FME to Solve Spatial and Non-Spatial Problems

24
Phil Tivel & Rich Seidlitz Great-Circle Technologies Spatial and Non-Spatial uses of FME

Transcript of Using FME to Solve Spatial and Non-Spatial Problems

Page 1: Using FME to Solve Spatial and Non-Spatial Problems

Phil Tivel & Rich SeidlitzGreat-Circle Technologies

Spatial and Non-Spatial uses of FME

Page 2: Using FME to Solve Spatial and Non-Spatial Problems

Great-Circle Technologies is a Big Data analytics small business focusing on multi-INT predictive analytics and visualization.

We bring a holistic approach to create solutions for hard problems.

GeoCyber, multi-lingual SMA, and semantic, geospatial, and behavioral analytics.

Our answer isn’t more bodies, it’s more innovation.

Page 3: Using FME to Solve Spatial and Non-Spatial Problems

No Puzzle is a Single Piece

Data are diverse and plentiful – how do we normalize them for analysis?

Page 4: Using FME to Solve Spatial and Non-Spatial Problems

No Puzzle is a Single Piece

Data are diverse and plentiful – how do we normalize them for analysis?

How can we automate normalization?

Page 5: Using FME to Solve Spatial and Non-Spatial Problems

No Puzzle is a Single Piece

Data are diverse and plentiful – how do we normalize them for analysis?

How can we automate normalization?

What if our data are distinctly different?

Geospatial, text, various INT data

Page 6: Using FME to Solve Spatial and Non-Spatial Problems

Problem: Raster collection and processing can be difficult and time consuming. It often needs to be collected, managed, clipped, and mosaicked.

Solution: Let FME do it all!

Spatial Example

Page 7: Using FME to Solve Spatial and Non-Spatial Problems

DRG (Digital Raster Graphic) index for the USA

Represents 89,111 DRGs! Each DRG is 1:24,000 topographic map of that grids area and has a weird name like o36075g8.

Page 8: Using FME to Solve Spatial and Non-Spatial Problems

Raw DRG Data

Bonus: The DRG data has marginalia that overlaps each other and is projected in UTMs.

Page 9: Using FME to Solve Spatial and Non-Spatial Problems

Raw DRG Data

For the state of Virginia (727 DRGs) How long would it takeyou to download, re-project, clip off marginalia, mosaic together,and clip again to the county? ….Hours?...Days?...Weeks?

Page 10: Using FME to Solve Spatial and Non-Spatial Problems

The FME Model

Page 11: Using FME to Solve Spatial and Non-Spatial Problems

The Results• The Data is Downloaded• Re-projected to GCS WGS1984• Then the Marginalia is clipped off

Page 12: Using FME to Solve Spatial and Non-Spatial Problems

The Results

Then the data is mosaicked to the chosen state leveland output as a single raster (.tif)

Page 13: Using FME to Solve Spatial and Non-Spatial Problems

The Results

The mosaicked state raster is then clipped into a rasterfor each county within the state

Page 14: Using FME to Solve Spatial and Non-Spatial Problems

The FME model is versatile. It is easy to incorporateother types of raster datasets or even vector datasets

Versatility

Page 15: Using FME to Solve Spatial and Non-Spatial Problems

Non-Spatial Example

Geospatial data without context is a map. But with context, it becomes a story.

Text documents can provide that context.

Page 16: Using FME to Solve Spatial and Non-Spatial Problems

The Original Data

Text that can consist of multiple languages and emoticons

Page 17: Using FME to Solve Spatial and Non-Spatial Problems

The Results

FME model creates seven more fields in the data. Each containing new linguistic information about the data.

Page 18: Using FME to Solve Spatial and Non-Spatial Problems

The Results

A closer look

Page 19: Using FME to Solve Spatial and Non-Spatial Problems

The Results

A closer look

Page 20: Using FME to Solve Spatial and Non-Spatial Problems

The FME Model

Python plays a large role in this model. The model starts by usingpython to create a regular expression to find all known emoticonsin the input Excel file.

Page 21: Using FME to Solve Spatial and Non-Spatial Problems

The FME Model

Regular Expression

Page 22: Using FME to Solve Spatial and Non-Spatial Problems

Linguistic information can be derived because of theCharacter Code Extractor Transformer

Custom Transformer

The FME Model

Page 23: Using FME to Solve Spatial and Non-Spatial Problems

The data can be easily separated out at this point based on different language statistics:

If it contains an emoticon or possible emoticon

If it has single language or multi language content

If the content has like languages

What can be done now

Page 24: Using FME to Solve Spatial and Non-Spatial Problems

Thank You!

Questions?

For more information contact:

Phil Tivel

[email protected]

Richard Seidlitz

[email protected]