Nice people thread part 3- Nice as pie

PasturesNew

This type of discussion, imho, never works well online as there end up being too many tangents. I will not be posting my opinions/help as I will now leave it with tomtom, who has much more experience than me, more recently, to a higher level. And anything I say might muddy the waters.

First step for MissKool to think of though is: get the table structure right. Once the table structure is spot on the rest is "easy peasy" as it all just flows and works right when you then try to do more/new things with it.

tomterm8

PasturesNew wrote: »

This type of discussion, imho, never works well online as there end up being too many tangents. I will not be posting my opinions/help as I will now leave it with tomtom, who has much more experience than me, more recently, to a higher level. And anything I say might muddy the waters.
.

It's not very recent experience, PasturesNew, about four or five years ago, and as I said I am very rusty. Your contributions so far have been helpful, concise, and have not muddied the waters at all, so I would like them to continue.

misskool

Can we have everyone contributing please because I have ZERO knowledge.

To make it easier, this is a targeted study so where there is something that in the real world might have lots of options, in this case, there is only 2,3,4

say diagnosis, the patient will have only stage 1, 1b, etc. Hospital is one of 5 centres, surgery etc is only one of 2. the ER, PR, FISH is positive or negative. These are not yet relevant at this stage.

viva: would hyperlinked excel be easier than a database? I just need a report that says Pat ID A001 has x samples from this date, y samples from this date and they are stored in location a1, b7 etc

vivatifosi

Missk, how much data are we talking here? How many lines per spreadsheet? Pat-ID is the unique identifier in each record yes? Is that what you mainly search by?

FWIW I don't think this is that difficult a database. Most of your data is in one form, all of the rest is linkable through a common ID (providing no data entry errors of course).

tomterm8

Well... does that mean all the info you need for your report is in table 2?

vivatifosi

tomterm8 wrote: »

Well... does that mean all the info you need for your report is in table 2?

I thought it was all it table 1! Confused:o

misskool

Yes, most of the report is table 2 and 3. For now. Later on, I need to do some statistical analysis from table 1 based on the results of table 5 (which is the study data to be generated)..

misskool

vivatifosi wrote: »

Missk, how much data are we talking here? How many lines per spreadsheet? Pat-ID is the unique identifier in each record yes? Is that what you mainly search by?

FWIW I don't think this is that difficult a database. Most of your data is in one form, all of the rest is linkable through a common ID (providing no data entry errors of course).

Pat-ID is the unique identifier yes. The searches might be based on location as well as patient ID.

At the moment, samples are being collated so that I know which ones I have, then they have to be organised into final places. Much of the samples arrived before I started work so no one has been marking them off and storing them before I arrived.

edit: sorry, there are 200+ for one study, some of them will have 100, some will be a bit smaller. Recruitment is always on overdrive as there are lots of dropouts and mortality

vivatifosi

missk I'm from an information rather than IT background like Pastures and tomterm (please correct me if I'm wrong there peeps) so come at this slightly differently. If the dataset is still small enough, which it sounds as though it is, at this stage I'd be wanting to make sure that the data is as standardised as possible before I do anything. A couple of weeks ago I was doing something similar myself, from Polish to English. A thousand or so records are easy to work with. 10,000+, that's when headaches start.

Get it all nice and clean. For example, when I've tidied up data in the past, you may have had the same organisation with different entries, for example: West Yorkshire Hospital, W Yorks Hosp, West Yorks Hospital.... If you are happy you've got this data formatted ok then you should be able to take it over fairly easily into a database. But before you do anything, you back it up, then back it up again... Getting back to the start point if you fu is critical. Double check that the data isn't being worked on elsewhere so that there's one final version. Oh, and write data entry standards or standardise the entry form (have a table of possibilities so nothing else can be entered) in the future.

PasturesNew

tomterm8 wrote: »

It's not very recent experience, PasturesNew, about four or five years ago, and as I said I am very rusty. Your contributions so far have been helpful, concise, and have not muddied the waters at all, so I would like them to continue.

aww

Well, I've not really "used" MS Access since about 2003-2006 or so, I've got a copy of 2003 and I just used it to keep my bookmarks so they were searchable (I had about 3000 bookmarks), but I've not even bothered using that to look things up for the past 5 years.

I'm fine with normalising and programming theory/suggestions as to how the system might work, but Access is really a "use it or lose it" type of skill, especially as the different versions all changed so much. Back in about 1997 I was doing server support for SQLServer, on NT4, but again, that was donkey's years ago - and I was flying by the seat of my pants even then

I did the OU Databases & Data Modelling, which was a Level 3 course.... in 1996.

That's the thing you see, always doing courses, never ever ever got chance to use anything I learnt.