Difference between revisions of "Data Warehouse"

From HiveTool
Jump to: navigation, search
Line 15: Line 15:
 
*Compared with limits and send alerts if the data exceeds limits.
 
*Compared with limits and send alerts if the data exceeds limits.
 
*Transactional data processed
 
*Transactional data processed
 +
 +
 +
The data needs to be:
  
 
*Structure the data so that it makes sense to the researcher.
 
*Structure the data so that it makes sense to the researcher.
Line 21: Line 24:
 
*Maintain data and conversion history.
 
*Maintain data and conversion history.
 
*Improve data quality by flagging and fixing bad data and assigning quality codes and descriptions.
 
*Improve data quality by flagging and fixing bad data and assigning quality codes and descriptions.
 
The data needs to be:
 
 
 
*partitioned into yearly or seasonal periods,
 
*partitioned into yearly or seasonal periods,
 
*summarized (daily weight changes),
 
*summarized (daily weight changes),

Revision as of 17:37, 11 June 2017

Since HiveTool is open source/open notebook, the entire primary record is publicly available online as it is recorded. Two databases are used: Input and Analysis.

Each hive sends in data every five minutes, 288 times a day, inserting over 100,000 rows a year into the Input Database. One thousand hives would generate 100 million rows per year. In addition to the measured data, there are external factors that need to be systematically and consistently documented. This metadata includes hive genetics, manipulations, mite treatments, data conversion and calibration formulas, etc.

Operational and Research Databases


As the data is moved from the Input Database to the Analysis Database it should:

  • cleaned up,
  • Converted to other measurement systems (lb <=> kg, Fahrenheit <=> Celsius).
  • Manipulation changes filtered out
  • Weather data checked if recent, else fetched.
  • Compared with limits and send alerts if the data exceeds limits.
  • Transactional data processed


The data needs to be:

  • Structure the data so that it makes sense to the researcher.
  • Structure the data to optimize query performance.
  • Make research and decision–support queries easier to write.
  • Maintain data and conversion history.
  • Improve data quality by flagging and fixing bad data and assigning quality codes and descriptions.
  • partitioned into yearly or seasonal periods,
  • summarized (daily weight changes),
  • cataloged and tied into the metadata (foundation type, hive orientation, mite treatment, etc.),
  • tracked and controlled with version control software,
  • released for use by researchers for data mining, online analytical processing, research and decision support.

Data Warehouse