Friday, December 8, 2017

GIS 5935: Lab 15- Dasymetric Mapping

Good afternoon UWF,

The final lab for the semester was a little difficult, but some interesting observations were made on population vs. habitable land using dasymetric mapping.  The lab required us to match up high school districts with census tracts and find the population of each high school district.  Area factors were taken into consideration such as water bodies, land cover, and imperviousness.  All of these factors help determine just where exactly a population lives within a certain boundary, because the boundary itself will include areas that a humn can't physcially live on like lakes and roads..

The lab makes reference to a reference dataset that was used to actually collect data for legit population statistics.  We used this on our lab to compare the results of our dasymetric mapping methods to determine population.  Once again a lot of statistics and formulas were used that I struggled to understand.  I really need to be able to focus more and try to understand what some of these mean, but based upon my work and social schedule, it has been hard to fit any more study time into my life than what I am already doing.  I am glad that this semester is over as it has been my toughest one to date.

-Matt Griggs

Sunday, December 3, 2017

GIS 5935: Lab 14- Spatial Data Aggregation

Good evening UWF,

This week's lab was challenging and fun at the same time.  Challenging in the sense that there was still some statistics works involved (which I am still not a fan of) and fun in the sense of looking at some spatial analysis formulas that I have not been exposed to before.

The main focus of the lab was on gerrymandering and how its effects are seen in county and congressional district data among certain states.  Congressional District data can appear to look odd on the map in terms of their boundaries.  Odd an unique shapes cover different parts of states for various reasons.  It is supposed to be illegal to draw these boundaries based upon racial profiling, but it is hard to argue that some of these areas are probably drawn to include unique sections of communities as a why to divide certain areas of a larger community as a whole.  Our lab looked at the shape of these districts and measured their compactness and community levels.  It is ideal to think of a community as a close-knit, compact area, whether it be large or small.  Congressional Districts should be drawn relative to population size, but also shaped in a more compact area to incorporate the cohesiveness of a community.  This lab helped identify some violators of compactness and community.  The state of North Carolina had a lot of these violators according to my analysis.  The screenshots below show the Congressional District with the worst compactness score and worst community score based upon my analysis.

-Matt Griggs

Sunday, November 26, 2017

GIS 5935: Lab 13- Effects of Scale

Good evening UWF,

This week's lab dealt with the effects of scale on some DEM data.  I felt that the Kienzle reading required for this week was a great supplemental guide to follow when completing this lab.  This lab required the user to create many comparison tables between different DEM datasets, but still allowed some creativity in how to show this data.  Sometimes a table was appropriate where other times a histogram or chart was just as good.  It is always nice to work with statistics simultaneously with GIS (especially if the statistics are easier to work with unlike previous labs).

I felt that a strong tie to ground truthing could be another supplemental aid with this type of work that we are doing with scale in lab assignments.  In the flowline section of the lab I pointed out a few different stream intersection sites and showed how far off from each other they were based upon the data and its scale.  Generalizations of stream shape can be easily made, but survey methods and ground truthing would be the best way at proving accuracy, regardless of scale.

-Matt Griggs

Tuesday, November 21, 2017

GIS 5935: Lab 12- Geographically Weighted Regression

Good afternoon UWF,

This weeks lab dealt with using Geographically Weighted Regression in ArcMap.  I fell confident in saying that after these past 3 labs, I will never want to deal with so many statistic-driven geoprocesses again.  I was easily confused when trying to execute a lot of the deliverables asked for in this lab.  I still really don't understand a lot of source material from the past few labs, as statistics and math subjects are ones that I struggle with retaining information.  I am a visual learner, which means I prefer to see map outputs to help me visualize information better.  Some of the maps/screenshots in this lab helped me see some trends in the data, so that was reassuring to complete, regardless if my data is right.

The lab directions ask me to explain the difference between OLS and GWR.  GWR is a local model for understanding your data and its statistics.  OLS is a global model.  A GWR model can be better for smaller areas since distance factors are considered with coefficient values, where in OLS they are not (constant coefficient values for every location).  Ideally this lab was supposed to show an improvement in my data statistics between OLS and GWR results, but it did not for me.  I am sure this was because of the errors generated from my selection of explanatory variables when setting up my data in the beginning.  I don't have a ton of free time to "explore" this data more to select more appropriate variables, so I just went with what I felt was appropriate.

I am glad that this series of labs revolving around statistics are complete because I have not enjoyed working with this type of data at all.  I do not work around these types of statistics at work so I found it very difficult to relate the lab assignments to work that I have traditionally done.  Maybe in the future I will run into a project that requires more of this type of analyses, but until then I am ready to move on and learn something else in this class.  I am just happy to complete this lab this week as last week I was unable to due to time constraints and a lack of understanding similar source material to this week.

-Matt Griggs

Tuesday, November 7, 2017

GIS 5935: Lab 10- Introductory Statistics, Correlation, and Bivariate Regression

Good evening UWF,

This week's lab dealt with some introductory analyses revolving around statistics in excel.  I have to admit that I don't know my way around excel too well.  I know some introductory stuff but not a lot of advanced format and editing techniques.  I think that this lab and subsequent labs will introduce me to some new functions of excel that I have not seen before so I am excited about that.  At the same time labs like this scare me because they are a little overwhelming.  There were a couple of questions that I could not come up with an answer for. 

One powerful statistic that I will try to apply to work is the weighted average.  I think that I can use this at work with a lot of stormwater inspection data in terms of population densities for areas inspected, etc. 

-Matt Griggs

Wednesday, November 1, 2017

GIS 5935: Lab 9- Accuracy of DEMs

Good morning UWF,

This week's lab dealt with analyzing the accuracy of Digital Elevation Models.  This two part lab dealt with looking at accuracy at the percentile and root mean square error level first, then focused on elevation differences and standard summary statistic measures second. 

Part one of this lab focused a lot on determining if bias was present in the sample elevation points of the lab data.  This metric to determine accuracy looks for a systematic pattern of error.  The summary table required for this part of the lab required a calculation of root mean square error.  This metric always produces a positive value, so it can make bias hard to detect sometimes.  A RMSE of zero or close to zero is considered to be very good, but my results did not come close to zero.  Therefore I think that there is plenty of bias in my sampling point data used.  This bias may be linked to the sampling design of the points, as they were clustered together mostly and not randomly stratified amongst the study area.

Part two focused more on the error metric of determining accuracy.  Bias was not near as prevalent in the data used because the sampling point pattern was much more stratified compared to part one data.  I was able to calculate the difference between true elevation values and three different interpolated surface elevation values.  A table was created to show these differences as well as some basic summary statistics between the elevation differences.

All in all, this lab was fairly enjoyable.  I liked that I got to use a new geoprocessing tool in ArcMap called Extract Multi Values to Points.  This tool adds raster elevation values to point data.  This tool was critical to be able to see all of the elevation differences needed to run statistics on for this lab.  The table deliverable for this lab requires to show Deliverable 2 from Part A.  This table is an overall accuracy assessment of the DEM used.

-Matt Griggs

Monday, October 23, 2017

GIS 5935: Lab 8- Surface Interpolation

Good afternoon UWF,

This week's lab dealt with analyzing water sampling point data via different surface interpolation methods.  I like how this lab gave us the ability to compare and contrast four different interpolation methods to see which one we thought was the best.  I feel very confident in saying that there will be varied opinions amongst our class as to which method we think is the best.  I think that is the beauty of using GIS, is that there are some many tools and geoprocesses to explore when dealing with visual data.  So many statistics can be derived that are mostly unknown to the casual GIS user, and some of those were brought to my attention during this lab.

I have never used Thiessen polygon, IDW, or Splining before in my previous coursework or throughout my career.  I always appreciate being exposed to new learning experiences in GIS, and I try to apply most of what I learn into my daily work when possible.

Without explaining too much about each interpolation method used, the main takeaway that I have from this lab is that evaluating your data pre and post interpolation is very important.  Some of the outputs can be skewed based upon existing data, because these interpolations produce results as is.  In our lab I deleted a sampling point during the splining step because an anomaly was being produced.  Upon completion of the deletion the output looked much more accurate and compared favorably to the other interpolation output methods.

The map deliverable required for this post is displayed below.  They are the screenshots of all four interpolation techniques used in this lab.



Regularized Spline

Tension Spline

-Matt Griggs