Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,790 --> 00:00:06,580
All right one last conceptual lecture before we get our hands dirty and actually create some table relationships
2
00:00:07,140 --> 00:00:14,140
and I really just wanna drive home this very important idea of relationships versus merging tables.
3
00:00:14,170 --> 00:00:17,910
Now I know some of you out there are thinking like this guy.
4
00:00:18,130 --> 00:00:24,520
Can I just merge my queries or use familiar functions like look up a related or index match to pull
5
00:00:24,520 --> 00:00:29,680
attributes into the fact table itself so that I have everything in one place.
6
00:00:29,740 --> 00:00:32,970
And the answer is yes technically you can.
7
00:00:32,980 --> 00:00:38,530
You can create a table that looks like this where you've got your original data your metrics your fact
8
00:00:38,530 --> 00:00:43,930
table fields on the left just like the one we've been looking at with a date a product ID and a quantity
9
00:00:43,930 --> 00:00:50,020
field and you could take those key columns and you could stitch in attributes from a calendar lookup
10
00:00:50,020 --> 00:00:52,680
table attributes from our Product table.
11
00:00:52,780 --> 00:00:59,890
And this could go on and on and on until you have hundreds of columns based on a handful of IDS or keys.
12
00:00:59,890 --> 00:01:03,720
And the thing is this is a totally normal habit to be in.
13
00:01:03,730 --> 00:01:07,280
This was a habit that I had as a longtime Excel user.
14
00:01:07,450 --> 00:01:13,840
Where is this instinct to want to always blend and stitch things together and force my tables into one
15
00:01:13,840 --> 00:01:14,560
place.
16
00:01:14,830 --> 00:01:16,500
And there's a good reason for it.
17
00:01:16,570 --> 00:01:24,400
Before these tools came along like power Korean power pivot Dax in Excel there were very few user friendly
18
00:01:24,400 --> 00:01:29,180
intuitive tools that would allow you to do this type of data modeling work.
19
00:01:29,200 --> 00:01:35,970
So as a result you were kind of forced to mash these different sources of data together into these Franken
20
00:01:35,980 --> 00:01:41,830
tables that contain all sorts of information because that was the only way you can analyze it like if
21
00:01:41,830 --> 00:01:47,710
you wanted to use a traditional pivot table for instance you had to point that pivot table to a single
22
00:01:47,710 --> 00:01:49,660
data source or a single table.
23
00:01:49,960 --> 00:01:54,730
And that often required using hundreds of thousands of look up or index match functions.
24
00:01:54,730 --> 00:01:57,910
To tie this information together with brute force.
25
00:01:58,210 --> 00:02:00,160
So bottom line.
26
00:02:00,160 --> 00:02:06,140
Sure you can do this mechanically it's possible but it's just really really inefficient.
27
00:02:06,400 --> 00:02:12,940
Like we talked about in that normalization lecture merging data like this creates a ton of redundant
28
00:02:12,970 --> 00:02:14,070
information.
29
00:02:14,290 --> 00:02:20,680
As a result it utilizes way more memory a lot more processing power than simply creating relationships
30
00:02:20,950 --> 00:02:23,820
between multiple small thin tables.
31
00:02:23,860 --> 00:02:30,130
So next time you feel that instinct to mash everything together to merge it to stitch it manually just
32
00:02:30,130 --> 00:02:30,740
say no.
3546
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.