# Custom Datasets Efficiency **URL:** \ **Category:** Ignition\ **Tags:** ignition81, perspective, scripting, database\ **Created:** [April 9, 2025, 3:10pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899 "2025-04-09T15:10:50Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![arice](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@arice](https://forum.inductiveautomation.com/u/arice)\ **Post date:** [April 9, 2025, 3:10pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/1 "2025-04-09T15:10:50Z")
I am curious as to the most efficient way to accomplish creating a custom dataset created from a UDT that comes from OPC. Currently I have created this script in a project library: ```python def createTable(dryer_data): # Renamed DryerType to dryer_data for clarity # Create Header header = ['Name','Actual', 'Set'] SetName = dryer_data['Set']['Name'] ActualName = dryer_data['Actual']['Name'] SetNumber = str(dryer_data['Set']['Number']) ActualNumber = str(dryer_data['Actual']['Number']) SetDryingTime = str(dryer_data['Set']['DryingTime']) ActualDryingTime =str( dryer_data['Actual']['DryingTime']) SetCoolingTime = str(dryer_data['Set']['CoolingTime']) ActualCoolingTime =str( dryer_data['Actual']['CoolingTime']) SetInletTemp = str(dryer_data['Set']['InletTemp']) ActualInletTemp =str( dryer_data['Actual']['InletTemp']) SetOutletTemp = str(dryer_data['Set']['OutletTemp']) ActualOutletTemp =str( dryer_data['Actual']['OutletTemp']) SetCoolingTemp = str(dryer_data['Set']['CoolingTemp']) ActualCoolingTemp =str( dryer_data['Actual']['CoolingTemp']) SetReversingTime = str(dryer_data['Set']['ReversingTime']) ActualReversingTime =str( dryer_data['Actual']['ReversingTime']) SetUnloadingTime = str(dryer_data['Set']['UnloadingTime']) ActualUnloadingTime =str( dryer_data['Actual']['UnloadingTime']) SetInfaredTemp = str(dryer_data['Set']['InfaredTemp']) ActualInfaredTemp =str( dryer_data['Actual']['InfaredTemp']) SetInfaredTime = str(dryer_data['Set']['InfaredTime']) ActualInfaredTime =str( dryer_data['Actual']['InfaredTime']) SetAfterDryingTime = str(dryer_data['Set']['AfterDryingTime']) ActualAfterDryingTime =str( dryer_data['Actual']['AfterDryingTime']) # Create Rows rows = [ ['Program Name', SetName, ActualName], ['Program Number', SetNumber, ActualNumber], ['Drying Time', SetDryingTime, ActualDryingTime], ['Cooling Time', SetCoolingTime, ActualCoolingTime], ['Inlet Temp', SetInletTemp, ActualInletTemp], ['Outlet Temp', SetOutletTemp, ActualOutletTemp], ['Cooling Temp', SetCoolingTemp, ActualCoolingTemp], ['Reversing Time', SetReversingTime, ActualReversingTime], ['Unloading Time', SetUnloadingTime, ActualUnloadingTime], ['Infared Temp', SetInfaredTemp, ActualInfaredTemp], ['Infared Time', SetInfaredTime, ActualInfaredTime], ['After Drying Time', SetAfterDryingTime, ActualAfterDryingTime] ] # Create PyDataset dataset = system.dataset.toDataSet(header, rows) # Write the dataset to a tag return dataset ``` I then call this through a Historical Tag binding on a table like this: ![image](https://us1.discourse-cdn.com/inductiveautomation/original/3X/9/8/980c5c03954531c91987fc3fb880a5100b70def5.png) I was wondering is this the most efficient way to do this or is there something better like a transaction group that could accomplish this. The idea is to have this visualized in realitime in perpsective and stored to a database in way that is easily read. Also keep in mind that in in this UDT there are a lot of different data types that I am converting to a string so that the table structure can look like this: ![image](https://us1.discourse-cdn.com/inductiveautomation/original/3X/c/0/c0e68989b1d602dd26e69e0b68698a70eba909da.png) ---
**Author:** ![paul-griffith](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/paul-griffith/32/37249_2.png) [@paul-griffith](https://forum.inductiveautomation.com/u/paul-griffith)\ **Post date:** [April 9, 2025, 3:21pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/2 "2025-04-09T15:21:33Z")
I'm very confused. What do you think your tag history binding is doing, currently? Because it's not actually used in your transform, meaning you're running the binding for literally no purpose. As a secondary tangent: don't explicitly import project library scripts (`from createData import createTable`); just use them directly (`createData.createTable(x)`). ---
**Author:** ![arice](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@arice](https://forum.inductiveautomation.com/u/arice)\ **Post date:** [April 9, 2025, 3:23pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/3 "2025-04-09T15:23:19Z")
Ok that makes sense would this be better served just calling this script from a gateway timer script and setting it to a dataset tag? ---
**Author:** ![arice](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@arice](https://forum.inductiveautomation.com/u/arice)\ **Post date:** [April 9, 2025, 3:24pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/4 "2025-04-09T15:24:43Z")
To be clear the only reason i used the tag history was to enable the polling so that it would be updated at a specified rate. ---
**Author:** ![paul-griffith](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/paul-griffith/32/37249_2.png) [@paul-griffith](https://forum.inductiveautomation.com/u/paul-griffith)\ **Post date:** [April 9, 2025, 3:31pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/5 "2025-04-09T15:31:36Z")
Okay, yeah, so don't do that. If you need a binding to poll, use an expression `now(updateRateMilliseconds)` - at least if you're throwing that away it's going to run in a few microseconds, and not require a full network IO hop to retrieve history data that's then immediately discarded. But all that's a tangent. Let's go back a step: You have a bunch of individual tags here that you're trying to assemble into a dataset for display on a view. That's fine. Are each of these individual tags supposed to be stored into a database? You almost certainly want them to be logged **individually** - not in some kind of pseudo dataset format. If you just need trends over time, that's a fine job for the tag historian. If you want **event driven** storage, such as "when I change X, do Y", then you **need** either transaction groups or to just script database updates. This is a totally separate problem from the live display in a table problem, and should be treated as such. The "live display of a sequence of values" problem is best handled with bindings, so that you don't have to do the polling yourself and can reuse our tag subscription mechanisms for efficiency. Unfortunately it's _slightly_ convoluted to do this - you need to configure an inner view to use for each row of the table, and provide it tag **paths** to use internally in indirect bindings: > [@Script to generate bindings](https://forum.inductiveautomation.com/t/script-to-generate-bindings/104816/4): > > Fran, thanks for the link to my post. I don't think it's what Jose requires as it won't update if any of the tags change value. I think the easiest option is, Create a simple coordinate view with a tagPath parameter. Add a Label Component to the view and create a tag binding on the label using the tagPath parameter. Render the table's value column as a view and select the view created above. Pass the tag path to the view instance. You should be able to generate the list of instances by scrip… ---
**Author:** ![pascal.fragnoud](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/pascal.fragnoud/32/24728_2.png) [@pascal.fragnoud](https://forum.inductiveautomation.com/u/pascal.fragnoud)\ **Post date:** [April 9, 2025, 3:56pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/6 "2025-04-09T15:56:14Z")
This is not gonna solve anything, but it will certainly make things easier: ```python def create_table(dryer_data): # Renamed DryerType to dryer_data for clarity key_map = { 'Program Name': "Name", 'Program Number': "Number", 'Drying Time': "DryingTime", 'Cooling Time': "CoolingTime", 'Inlet Temp': "InletTemp", 'Outlet Temp': "OutletTemp", 'Cooling Temp': "CoolingTemp", 'Reversing Time': "ReversingTime", 'Unloading Time': "UnloadingTime", 'Infared Temp': "InfaredTemp", 'Infared Time': "InfaredTime", 'After Drying Time': "AfterDryingTime" } data = [ [k, dryer_data['Set'][v], dryer_data['Actual'][v]] for k, v in key_map.iteritems() ] return system.dataset.toDataSet( ['Name','Actual', 'Set'], data ) ``` - This is without assuming anything about the data coming in. - Keep your variables names case consistent: You're mixing snake\_case (`dryer_data`) and PamelCase (`SetNumber`). Ignition uses camelCase. PEP8 recommends snake\_case for function and variable names. PascalCase is reserved for classes. - `# Create Rows`: please don't. Comments need to be useful, or they only add clutter and make things less readable. ---
**Author:** ![arice](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@arice](https://forum.inductiveautomation.com/u/arice)\ **Post date:** [April 9, 2025, 3:57pm UTC](https://forum.inductiveautomation.com/t/custom-datasets-efficiency/104899/7 "2025-04-09T15:57:37Z")
Ok I think I understand what you are saying. These are two seperate issues. For storing the tags to the database I should connect the UDT individual tags to a historian. Then for the live view I should create the table as custom component that I can pass in the UDT as a parameter and will connect to each of the inner views rows like this: ![image](https://us1.discourse-cdn.com/inductiveautomation/original/3X/1/f/1f06bd6e93d2c67a14a7a9fb7eb2528a736908bb.png)