We recently received a number of false/bogus reports on our hotline. Does anyone have experience or thoughts about how to process and ultimately determine a resolution for these kinds of reports that don't skew your year-end numbers and reporting?
We recently received a number of false/bogus reports on our hotline. Does anyone have experience or thoughts about how to process and ultimately determine a resolution for these kinds of reports that don't skew your year-end numbers and reporting?
1
Ernesto Gonzales
commented
Oct. 19, 2023, 3:54 p.m.
Hi Debra - This is a great question. Here's my short answer: outliers need to be addressed during the Exploratory Data Analysis stage (when they are found) and wrangled/transformed once it is known more about them.
There will be instances that outliers need to be part of a report, and also when they need to be removed, standardized or replaced.
Filtering them out or replacing them with the average value of the data set are some of the ways to address outliers.
Data needs to be wrangled, cleansed, and formatted before doing a report. There are guidelines online about how and when to remove/filter outliers.
Meanwhile, an outlier might be an indicator of a new trend in a data, or a potential insight as well. It depends on the context. It all starts at the data source/input. User error, unusual circumstances, and data integrity, to name a few.
Hope this helps.
Ernesto Gonzales
commented
Oct. 19, 2023, 3:54 p.m.
Hi Debra - This is a great question. Here's my short answer: outliers need to be addressed during the Exploratory Data Analysis stage (when they are found) and wrangled/transformed once it is known more about them.
There will be instances that outliers need to be part of a report, and also when they need to be removed, standardized or replaced.
Filtering them out or replacing them with the average value of the data set are some of the ways to address outliers.
Data needs to be wrangled, cleansed, and formatted before doing a report. There are guidelines online about how and when to remove/filter outliers.
Meanwhile, an outlier might be an indicator of a new trend in a data, or a potential insight as well. It depends on the context. It all starts at the data source/input. User error, unusual circumstances, and data integrity, to name a few.
Hope this helps.
