04-08-2020
MikeJenista
SAS Employee
Member since
08-03-2016
- 32 Posts
- 0 Likes Given
- 6 Solutions
- 3 Likes Received
-
Latest posts by MikeJenista
Subject Views Posted 1051 04-08-2020 08:39 AM 3989 07-16-2018 09:17 AM 4102 07-09-2018 03:57 PM 4110 07-09-2018 02:37 PM 4117 07-09-2018 02:29 PM 1598 07-09-2018 02:24 PM 1606 07-09-2018 02:09 PM 4127 07-09-2018 02:00 PM 1922 07-09-2018 01:23 PM 1841 07-09-2018 01:03 PM -
Activity Feed for MikeJenista
- Got a Like for Re: SAS VISUAL ANALYTICS 8.2. 04-08-2020 10:27 AM
- Posted Re: SAS VISUAL ANALYTICS 8.2 on SAS Visual Analytics. 04-08-2020 08:39 AM
- Got a Like for Re: SAS Visual Statistics and Visual Analytics. 07-17-2018 03:12 AM
- Posted Re: Creating datasets on the fly - SAS VA on SAS Visual Analytics. 07-16-2018 09:17 AM
- Got a Like for Re: Creating datasets on the fly - SAS VA. 07-10-2018 05:37 AM
- Posted Re: Creating datasets on the fly - SAS VA on SAS Visual Analytics. 07-09-2018 03:57 PM
- Posted Re: Creating datasets on the fly - SAS VA on SAS Visual Analytics. 07-09-2018 02:37 PM
- Posted Re: Creating datasets on the fly - SAS VA on SAS Visual Analytics. 07-09-2018 02:29 PM
- Posted Re: how to create group by average column in SAS VA 7.3 crosstab on SAS Visual Analytics. 07-09-2018 02:24 PM
- Posted Re: SAS VISUAL ANALYTICS 8.2 on SAS Visual Analytics. 07-09-2018 02:09 PM
- Posted Re: Creating datasets on the fly - SAS VA on SAS Visual Analytics. 07-09-2018 02:00 PM
- Posted Re: MoM growth of the current month in text object on SAS Visual Analytics. 07-09-2018 01:23 PM
- Posted Re: MoM growth of the current month in text object on SAS Visual Analytics. 07-09-2018 01:03 PM
- Posted Re: Sankey Diagram -- Sorting variables on SAS Visual Analytics. 06-27-2018 09:28 AM
- Posted Re: sas va calculate rate between first N revenue and total revenue on SAS Visual Analytics. 06-27-2018 09:16 AM
- Posted Re: regarding List table on SAS Visual Analytics. 06-26-2018 12:03 PM
- Posted Re: Trouble keeping date format when exporting to Excel on SAS Visual Analytics. 06-26-2018 11:59 AM
- Posted Re: Timeseries graph question on SAS Visual Analytics. 06-26-2018 11:29 AM
- Posted Re: How to change the date expression to numeric after creating 'New Calculated Item' in SAS VA on SAS Visual Analytics. 06-26-2018 11:16 AM
- Posted Re: How to change the date expression to numeric after creating 'New Calculated Item' in SAS VA on SAS Visual Analytics. 06-26-2018 11:14 AM
-
My Liked Posts
Subject Likes Posted 1 04-08-2020 08:39 AM 1 06-26-2018 10:52 AM 1 07-09-2018 03:57 PM
04-08-2020
08:39 AM
1 Like
I have had this problem as well--the paths displayed in a lot of CAS examples are paths relative to the where the session was started. It can be difficult to find content like code files through other file navigations. There is an additional complication now that CAS libraries have to be defined with relative paths for security reasons. I am not aware of a way to ask a CAS session for the current working directory, or equivalent. The way I got around this in the past was to ask the system administrators about the directory structure, and they usually knew how to reconcile relative CAS locations with the absolute machine locations.
... View more
07-16-2018
09:17 AM
The latest version of VA on Viya does have more flexible aggregation features. With the use of an aggregated temp table, users can write expressions similar to the Qlikview code you presented.
... View more
07-09-2018
03:57 PM
1 Like
Okay, I see that the problem is that the quartile/median aggregations need to have the aggregated AHT values available for ranking, so there is no way to figure out the quartile cutoffs from row-level data alone.
You can create an order-able object like a list table, and rank all the employees by AHT. But I don't think it is possible to automatically extract the cutoffs from this either. If the number of calls and talk time were already condensed for each employee, this would work since AHT would be a row-level calculation and not an aggregation. But with the need to aggregate over the time period, I don't think you can dynamically obtain the bucketing you want.
... View more
07-09-2018
02:37 PM
What version are you using? I can try setting up a similar scenario to see what I can figure out.
... View more
07-09-2018
02:29 PM
Okay, I think there is a possible way to do this. You can create three aggregated items, 'Threshold Q1', 'Threshold Q2', 'Threshold Q3'. These would just aggregate AHT, but instead of Sum or Average you would define each one to use a different quartile cutoff (Q1, median, Q3 I think are the correct options).
Then you can create a new calculated item something like this:
if AHT < Threshold Q1, then return 'Q1'
else if AHT < Threshold Q2, then return 'Q2'
else if AHT < Threshold Q3, then return 'Q3'
else return 'Q4'
... View more
07-09-2018
02:24 PM
The problem is that you can't assign a fixed scope for the aggregation, so when you show the Agent column, you are getting the 'average' value for each Manager-Agent combination (i.e. that agent's sales).
There is a new feature in the latest Viya versions that allows the scope of the aggregation to be manipulated, but at 7.3 I think you might be able to get what you want with a hierarchy. If you define a hierarchy on Manager->Agent, then your crosstab will have dynamically-expanding rows. When Agent is collapsed, you will see the average sales for the manager's agents. When the Agent is expanded, the Average Sales will show 'wrong' in the rows, but the subtotal for the Average Sales should be the what you want.
If you don't need the total sales for the manager, you could just change the aggregation type of 'Sales' to be 'Average' and you get the same effect with a single item.
... View more
07-09-2018
02:09 PM
1) Not within VA; you have to export the model and then you can change the score code. For models that use an ASTORE table for the scoring process, I don't think you can manipulate the model in any detailed way.
2) No; you would need to use a dedicated forecasting product for that (e.g. Forecast Studio).
3) Do you mean create a new calculated column? This is available from the Data panel.
4) There are some limits, usually on the number of variables that you can assign to a model. Rows should not be limited, although Forecasting does require time binning to avoid plotting too many points.
5) I don't think so. The data exports as a nice Excel sheet, but the charts/graphs download as images or PDFs in a separate feature (export data versus 'print'). The print feature does also print the details tables, which contain the model parameters.
... View more
07-09-2018
02:00 PM
Can you be more precise about what you want the quartile buckets to do? Are you trying to calculate the AHT for each employee, and then bucket their values by quartile?
Or are you trying to calculate the AHT in four quartiles of employees (bucketed in some other way)?
... View more
07-09-2018
01:23 PM
Okay, the simplest way I know to get the growth number to show is to assign the aggregated item to a chart where it would show as a single element. For example, if you use the same rank you describe on a pie chart, you would get the aggregated value for only one slice (showing as the whole wheel). You could then use the style and properties tabs to make the numeric value show very large and provide a title.
... View more
07-09-2018
01:03 PM
What version of VA are you on? The latest versions have a KPI object that does exactly what you want, but for earlier versions there are still some possibilities.
... View more
06-27-2018
09:28 AM
This is a late reply, but I wanted to make sure you got a complete response.
The features you desire are not supported on Sankey diagrams.
Sankey uses the natural sort for the event. Explorer does not have custom sorting, I think that is only for reports in Designer. At VA 8.2 and higher on Viya, all these features have been merged so it would be possible.
The time variable is not shown because it is only used to determine the sequence of each transaction. Two different transactions with the same sequence but differing times will be merged into the same path. Therefore, the time is not necessarily unique for any particular event shown at a stage in the Sankey diagram.
You could try the vertical layout option; that would put 'Bottom 99%' on the left side of the graph and 'Top 1%' on the right side, so left-to-right you have an increasing order.
... View more
06-27-2018
09:16 AM
There are a couple ways to get what you want, but I think the most flexible way is to create a parameter 'topN_cutoff' and a calculation 'topN_revenue' like:
if 'revenue'n > 'topN_cutoff'
return 'revenue'n
else 0
Then you can create an aggregated calculation like this:
sum [_ForAll_] ('topN_revenue'n) / sum [_ForAll_] ( 'revenue'n )
The only sticky point is that you need to know what the cutoff revenue is beforehand. For any given data, it is easy to find it out with a list table and a rank applied for top 3 (or whatever rank). But I don't think VA has an operator you can use to determine it automatically. The closest would be the Percentile operator which gives you the revenue value in the data for a specified percentile, but that won't correspond to a specific numeric rank (unless, again, you already know the data and can just determine what percentile corresponds with top N).
... View more
06-26-2018
12:03 PM
Do you mean you have a column called 'name' and you want to sort the table by its values? Or are you trying to sort the columns by name (I don't think that is possible in VA)?
If just sorting on a single column, it is possible that it is working but taking a very long time to return the result. How big is the table/data source?
... View more
06-26-2018
11:59 AM
Just a quick sanity check: did you try expanding the excel column with the ### values in it? I have sometimes had excel represent values that way until I expand the column width enough to show the full values.
... View more
06-26-2018
11:29 AM
Create a calculated item on the time variable that just gets the date ( DatePart( YourTimeVariable ) ).
Then define your 'Day' filter on that calculation and it should line up better. If it doesn't work, I think you might have a time-zone issue/bug?
... View more