Have recently been shown the full capabilities of QLIK and excited to introduce it into the business so thought I would have a crack at the free trial and see how in depth I am able to go.
This question I feel has most likely an incredibly simplistic answer and once this hill is passed a lot of forum posts already have answers for next steps to come.
Trialing QLIK I am importing a excel doc just to get a basis of data inconsistency in our system and have seen another user be able to combine multiple different entries under the one name. E.g. we could have customers with name discrepancies like Alex White and also entered as Alexander White. In QLIK these show as two separate entries and I cannot seem to combine them under the one so I can view total sales under the one name.
Can this be done within QLIK or is this a data cleanse required in the excel document to be loaded into QLIK? Thank you in advance for your response!
Such different namings could be unified per mapping (for the howto take a look on the help) maybe in combination with various string-functions like trim/lower/upper and so on. But the more differences are possible the higher the needed efforts to clean/repair the data - and it would be done within a successor system and not within the previous systems respectively the source. If any possible I wouldn't work with names else with (numerical) ID's.
Unfortunately due to the different softwares we do use the feasibility of introducing numerical IDs would be incredibly challenging as this would most likely lead to further human error.
In the preview I did have of QLIK the user had managed to data match within QLIK itself whereby a master name was generated and those which had similar entries or shared address or a form of matching variable were able to be listed under this one master name. In some cases one customer did have 4 different entries which could be viewed by selecting the master name in the list and this would then allow viewership of the matched names to further assist with tracing that entry back to its original data source. Will keep trying to look!
Like above mentioned it's possible to unify a differently naming with the help of various string-functions and/or a mapping-approach.
To implement a proper data-management is surely not an easy task but it's far away from impossible - it's just a decision to do now the preparations to implement it in x months. Only the fact that it couldn't be changed at once doesn't mean that it couldn't be done or isn't important enough because until now it works in some way ... without such change it will remain error-prone and there won't be a real control over the data - especially if the amount of data and/or their meaning increased (quite probably it will) you may be able to handle one or two poor processes but if the poor processes goes in the dozens ...