I applied some data analysis concepts to a dataset obtained from Tau Beta Pi HQ.
- gain a better understanding of the Texas Gulf Coast Alumni Chapter of Tau Beta Pi’s constituency
- determine parts of town where events might be hosted to achieve greater turnout
Data was received in .XLSX format, so I naturally gravitated towards Microsoft Excel for data analysis.
I constructed a series of PivotTables to gain insight about the geographical distribution of the alumni chapter’s constituency.
Prior to mapping, I filtered to count only those entries that have an email address, yielding 2682 out of 9127 records. My focus is on those members that have provided their email addresses: without external support, it will be challenging and expensive to reach out to the broader population.
I discovered OpenHeatMap (GitHub) while searching for a tool that would easily map entries from a CSV file. Thanks to Pete Warden for producing OpenHeatMap.
Tested the effect of cleaning ZIP code data, stripping any ZIP+4 codes back to five-digit ZIP codes.
- Some members list their work address instead of their home address, resulting in inflated counts from some parts of the city (namely downtown and Rice University)
- OpenHeatMap did not correctly resolve all ZIP codes
- This was noticeably worse when mapping ZIP+4 (ZIP+4 openheatmap.com vs. ZIP openheatmap.com)
I attempted to overcome the second limitation by providing more contextual location information, appending city and state to the ZIP code.
Note that the maps above were generated prior any data cleaning.
The final map was generated after cleaning the data of any misspelled cities
20170222 I learned of Google FusionTables, which can be used to generate heatmaps and much more. Beginning further analysis to see how this tool might be used. So far I have observed the Google Maps Geocoding service used by FusionTables to be rather slow – perhaps it’s single-threaded?
Unfortunately, the heatmaps generated by Google FusionTables show only 1000 rows of data. Additional limitations are listed here.
My Cool ‘n’ Quiet workstation was working fine until some point last night where it showed a black screen with moveable cursor. Attempts to bring up the Task Manager were unsuccessful – the cursor briefly changed to indicate busy state, but Task Manager would not appear.
This morning, the system would boot, displaying the Windows logo, but I was stuck on a black screen showing only my cursor.
I did not make any changes to hardware, and was not aware of any changes made to my system.
I plugged my secondary display into my workstation’s graphics card and attempted to tinker with Windows 10’s Project settings (Windows key + P) to no avail.
Stumbling Across a Solution
I poked around within my workstation’s UEFI and focused on Boot Option Priorities.
Last Known Good Configuration
Boot Option Priorities
#1 Windows Boot Manager
#2 SATA: My Windows 10 OS Drive
#3: UEFI: Built-in EFI Shell
I modified Boot Option Priorities:
- Swapping positions of #2 and #1 resulted in failure to boot
- Tested UEFI, Windows Boot Manager, and finally SATA. This time, I was greeted by the familiar login screen
Upon login, Windows displayed a notification stating that updates were installed.
Update history (Settings > Update & security > Windows Update > Update History) showed that an updated NVIDIA display driver (NVIDIA – Display – 12/29/2016 12:00:00AM – 126.96.36.19953) was installed last night – could this have been the culprit?
Confirmed that this updated NVIDIA display driver is shown for my graphics card within Device Manager.
After successful boot, I restarted my workstation, and reverted changes to Boot Option Priorities. I was still able to boot cleanly into Windows 10.
I ultimately modified Boot Option Priorities. They are now set to
- Windows Boot Manager