I am new to DynamoDB so I'm still trying to understand how to use it, but I have what I believe is a simple task but I'm not sure how to address it.
I need to create a table to store categorized questions in which I need to store a click counter. So let's say something like this:
ID: 1
Question: What is this?
Category: General
Clicks: 100
Now, the problem is I need an optimized way to get the most general clicked questions and the most clicked questions by category, let's say a top 10.
In a classic SQL style it would be something like this:
SELECT ID, Question
FROM Questions
ORDER BY Clicks DESC
LIMIT 10
Can anyone point me in the right direction on how to structure the table? I tried the sorting but it always requires a hash key condition, so I don't understand how I can get this done as I need the top 10 results and not a single one.
Thanks in advance!
How are you accumulating the clicks, if you are able to figure out how you accumulate the clicksstream onto the table correctly that will be your answer.
You will need to implement a mechanism that maps incoming clicks to the item record being clicked on and increment it using an atomic counter. With this you will be able to then create a sparse index and sort it in descending order to get what you need.
Related
I created a data table containing several columns: most of them must show the total of their respective values. I’m just wondering if there’s a way to selectively hide this sum, since the latest column includes an average of ratings and it doesn’t make sense to have a sum of them.
Below, an example of what I’d like to achieve if it’s possibile:
Basically, I need to hide the number deleted in red. I know that there could be plugins to do so, but I don’t know if I can ask to install one or more.
Thanks in advance.
Right now, I am using a filtering scheme which only looks at the data of the 5 or 6 most common entries in the 'Clinic' field. But, there are a handful of other possibilities which might account for a few rows each. They are too inconsequential to include on their own (I am using pie charts and bar charts), but I would like these rows to be accounted for. For this reason, I would like to create an "Other" category which groups these entries together. What is the best way of doing this? I know I can create a calculated column that groups everything aside from the top 5 or 6 in an other category, but I thought there might be a way to keep working with the original column and achieve the same result.
Unfortunately not. In 6.5.x you will have to write a case statement that will specify everything that is not most common to other.
In 7.0.x you can go to insert binned column. Add the bottom you can use values to create a bin. Add the values you want to the bin and call them "Other". Of course if you look at the column created like this, it is a case statement. But it is a whole lot faster than writing it yourself.
Following phiver I came out with this solution in Spotfire 6.5.2:
Add a calculated column
With something like this:
If(DenseRank(Count([Formation]) OVER ([Formation]),"desc")<10, DenseRank(Count([Formation]) OVER ([Formation]),"desc") & " " & [Formation], "10 Other")
Hope that helps.
I'm testing Kibana 4 for a project.
I have created an index from my database table which is composed by 3 fields:
Date
User
Action
I would like to display my index as a simple table (3 column, N rows) in my dashboard.
I tried to use "Data table" visualization but I can't find a way to display my results without any Metrics (Count, Sum etc...)
Maybe is pretty simple and I missed something... is there a way to do this?
Regards,
On the Discover tab, create a view that has just the fields you want and then save that as a search.
On the Dashboard tab, click on Edit then hit the + Create new button to add a widget, but if you look at the top, there's a Searches tab. Select that and add your saved search in.
[Elastic 7.x / 2019 Update]
I was a bit confused when I read #Alcanzar's answer so I am sharing a little more noob-friendly step-by-step how-to here :
STEP 1 : Create the Index Pattern
STEP 2 : Go to the Dashboard view, and create a view on your index
Select each column you want to include/add in your view by clicking "add" on it (The confusing part is that until you do that, you will have a "scrambled" view listing everything in a jumbled way.)
STEP 3 : Go to the Dashboard view, and create a view on your index
The trick is to select the specific columns you want to include... and voila !
Don't forget to save your view, this will help a lot in the process.
In Kibana 7.5.0 you can do it as follows:
Go to Discover section
Select fields you are interested in
Click on Save to save your discover search so you can use it in visualizations and dashboards
Click on Dashboard and create a new dashboard
Click on Add and select the panel
There is no step 6
The accepted solution has its pros (if, for simplicity, you see your index as a table, this is the only way to deal with rows naturally) but also cons (it allows the user to see too much information, by expanding the records that appear in the table; users cannot get an export of the values).
So if you plan to build tables to use in reports seen by users which should not see everthing and may want to get exports of the data, I recommend a different (hacky) approach using Table visualizations:
Say you have three columns A, B and C:
If there are no duplicates considering the combined values of A and B, you can use these two vales as aggregation fields, and then set a Max or Top hit Metric for C.
If even A, B and C have duplicates, then you can use the three of them as aggregation fields and add a Metric count, that will give you the number of repeated rows. This solution makes somehow sense, because instead of repeating the same row 'n' times you just tells you should have repeated 'n' times that row.
If A and B have duplicates but A, B and C are unique, then there is, afaik, no elegant solution. You have to use the three of them as aggregation fields, but then you would have a dummy metric at the end (e.g. count, always equal to 1).
Why? why do we have to go through all of this? that is another question...
I'm testing Kibana 4 for a project.
I have created an index from my database table which is composed by 3 fields:
Date
User
Action
I would like to display my index as a simple table (3 column, N rows) in my dashboard.
I tried to use "Data table" visualization but I can't find a way to display my results without any Metrics (Count, Sum etc...)
Maybe is pretty simple and I missed something... is there a way to do this?
Regards,
On the Discover tab, create a view that has just the fields you want and then save that as a search.
On the Dashboard tab, click on Edit then hit the + Create new button to add a widget, but if you look at the top, there's a Searches tab. Select that and add your saved search in.
[Elastic 7.x / 2019 Update]
I was a bit confused when I read #Alcanzar's answer so I am sharing a little more noob-friendly step-by-step how-to here :
STEP 1 : Create the Index Pattern
STEP 2 : Go to the Dashboard view, and create a view on your index
Select each column you want to include/add in your view by clicking "add" on it (The confusing part is that until you do that, you will have a "scrambled" view listing everything in a jumbled way.)
STEP 3 : Go to the Dashboard view, and create a view on your index
The trick is to select the specific columns you want to include... and voila !
Don't forget to save your view, this will help a lot in the process.
In Kibana 7.5.0 you can do it as follows:
Go to Discover section
Select fields you are interested in
Click on Save to save your discover search so you can use it in visualizations and dashboards
Click on Dashboard and create a new dashboard
Click on Add and select the panel
There is no step 6
The accepted solution has its pros (if, for simplicity, you see your index as a table, this is the only way to deal with rows naturally) but also cons (it allows the user to see too much information, by expanding the records that appear in the table; users cannot get an export of the values).
So if you plan to build tables to use in reports seen by users which should not see everthing and may want to get exports of the data, I recommend a different (hacky) approach using Table visualizations:
Say you have three columns A, B and C:
If there are no duplicates considering the combined values of A and B, you can use these two vales as aggregation fields, and then set a Max or Top hit Metric for C.
If even A, B and C have duplicates, then you can use the three of them as aggregation fields and add a Metric count, that will give you the number of repeated rows. This solution makes somehow sense, because instead of repeating the same row 'n' times you just tells you should have repeated 'n' times that row.
If A and B have duplicates but A, B and C are unique, then there is, afaik, no elegant solution. You have to use the three of them as aggregation fields, but then you would have a dummy metric at the end (e.g. count, always equal to 1).
Why? why do we have to go through all of this? that is another question...
Anyone ever see how fogbugz sorts their tables? When you click to sort the column, they actually break the table up into many small tables that have each category of info.
Wondering if anyone knows how they do this?
Looking to implement this feature.
If you take a look through the cases page, and sort you can see what I mean.
Any help would be AWESOME!
Still Haven't figured this one out.
EDIT: #Peter, I don't want to postback and recreate a table every time the header title is clicked for a sort. I also want to know if their is a generic solution for this. If I click on the header to sort, by the way of javascript, it seperates the "one" table into many and I want to know if their is any generic solution for this because its just a MUCH better way of viewing a sorted Table.
EDIT: I do need a javascript sorter, but if you look right down at the implementation of fogbugz, it produces a different result...
Yup, Rich got it (I coded this feature into FogBugz a long while back).
If you have to do this on the client you have no choice but to sort the data, iterate through it generating table row after table row, and every time you hit a new sort value you create a new thead w/ the appropriate information.
To be honest it would be a pretty cool modification to this jQuery plugin: http://tablesorter.com/docs/ and you'd be able to leverage a lot of their work. If you're going to put in the time and create a general solution, might as well make it accessible to the community.
Without knowing specifically how Fog Creek accomplishes this, the way that I would do it is to output a table header, then iterate through the list, outputting a footer and a new header each time the group value changed.
Not sure what answer do you expect. SQL query for this would simply use ordering on selected column, and UI would start new table each time this value changes.
Here is screenshot of FogBugz with this sorting, after clicking on Priority column.
http://img297.imageshack.us/img297/6974/76755363ee3.png
Of course, starting new table doesn't make sense for every column (title, case #).
Edit: If I understand correctly, you're looking for a way how to do this in a browser without loading new page. If this is the case, I would suggest at least some server-side support, which would return your data in correct order, and properly structured for subtables (in xml/json/whatever you use). Your javascript will use this data to recreate tables. I am sure others with more web-ui experience will provide you with better answers.
I've used the Sortable Tables script from Kryogenix with some good results.
I don't know if it is relevant, but we store the results of a query in a temporary table in SQL, and then reference current-row-less-one to see if a Category has changed, and indicate this in the resulset.
In some instances we "indicate" this with a column containing
<tr><td colspan=999>Category Heading</td></tr>
so that the web page can just "inject" that into the table it is building.
SELECT Col1, Col2, ...,
[CATEGORY] = CASE WHEN T1.CategoryCol <> COALESCE(T2.CategoryCol, '')
THEN '<tr><td colspan=999>' + T1.CategoryCol + '</td></tr>'
ELSE ''
END
FROM #MyTempTable AS T1
LEFT OUTER JOIN #MyTempTable AS T2
ON T2.ID = T1.ID - 1