Data Warehousing - Star Schema vs Flat Table - data-warehouse

I'm trying to design a Data Warehouse for a single store of commonly required data ranging from finance systems, project scheduling systems and a myriad of scientific systems. I.e. many different data marts.
I have been reading up on Data Warehousing and popular methods such as Star Schemas and Kimball methods etc but one question I cannot find answer to is:
Why is it better to design your DW Data Mart as a star schema rather than a single flat table?
Surely having no joins between facts and attributes/dimensions is faster and simpler than having lots of small joins to all the dimension tables? Disk space is not a problem, we'll just throw more disks at the database if necessary. Is the star schema slightly outdated these days or is it still data architect dogma?

Your question is very good: the Kimball mantra for dimensional modelling is to improve performance and to improve usability.
But I don't think it is outdated, or dogma- it is a reasonable, practical approach for many situations and platforms.
The way relational DBs store data means there's a balancing act to be struck between the numbers and types of tables, the routes in to the data for typical queries, easy maintainability and description of relationships between data, the numbers of joins, the way the joins are constructed, the indexability of columns, etc.
3NF (or further) is one end of the spectrum, suiting OLTP systems, and a single table is the other end of the spectrum. Dimensional models are in the middle and appropriate for reporting, at least when using certain technologies.
Performance isn't all about 'number of joins', although a star schema performs better for reporting workloads than a fully normalised database, in part because of a reduce number of joins. Dimensions are typically very wide. If you are including all those dimension fields in every row of every fact, you have very large rows indeed, and finding your way into those rows will perform very badly for typical queries.
Facts are numerous, so if you can make those tables compact, with the 'wordier' dimensions filterable, you hit a sweet spot of performance that a single table isn't going to match, unless heavily indexed.
And yes a single table for a fact is simpler in terms of numbers of tables but is it really easier to navigate? Dimensions and facts are easy concepts to understand, and what if you want to cross you queries across facts? You've got many different data marts but one of the benefits of having a data warehouse in the first place is that these aren't distinct- they're related and can be reported across. Conformed dimensions enable this.

If you combine your fact and dimensions into a single table, you'll either lose the visibility into dimension attributes that have never been used, or your measures will be thrown off by inclusion of a dummy event for the unused dimension attribute.
For example, a restaurant menu is a dimension and the purchased food is a fact. If you combined these into one table, how would you identify which food has never been ordered? For that matter, prior to your first order, how would you identify what food was available on the menu?
The dimension represents the possibilities, the fact represents the realization of the possibilities.

Combining facts and dimensions in the same table limits the scalability and the flexibility.
Suppose that one day the business decides to change a dimension description ( for example the product name ). Dimension tables aren't as deep as the fact tables and the update process or SCD management should be easier and less resource intensive.

Related

How to identify relevant columns in very wide tables using AI and Machine Learning?

I have a complex data model consisting of around hundred tables containing business data. Some tables are very wide, up to four hundred columns. Columns can have various data types - integers, decimals, text, dates etc. I'm looking for a way to identify relevant / important information stored in these tables.
I fully understand that business knowledge is essential to correctly process a data model. What I'm looking for are some strategies to pre-process tables and identify columns that should be taken to later stage where analysts will actually look into it. For example, I could use data profiling and statistics to find and exclude columns that don't have any data at all. Or maybe all records have the same value. This way I could potentially eliminate 30% of fields. However, I'm interested in exploring how AI and Machine Learning techniques could be used to identify important columns, hoping I could identify around 80% of relevant data. I'm aware, that relevant information will depend on the questions I want to ask. But even then, I hope I could narrow the columns to simplify the manual assesment taking place in the next stage.
Could anyone provide some guidance on how to use AI and Machine Learning to identify relevant columns in such wide tables? What strategies and techniques can be used to pre-process tables and identify columns that should be taken to the next stage?
Any help or guidance would be greatly appreciated. Thank you.
F.
The most common approach I've seen to evaluate the analytical utility of columns is the correlation method. This would tell you if there is a relationship (positive or negative) among specific column pairs. In my experience you'll be able to more easily build analysis outputs when columns are correlated - although, these analyses may not always be the most accurate.
However, before you even do that, like you indicate, you would probably need to narrow down your list of columns using much simpler methods. For example, you could surely eliminate a whole bunch of columns based on datatype and basic count statistics.
Less common analytic data types (ids, blobs, binary, etc) can probably be excluded first, followed by running simple COUNT(Distinct(ColName)), and Count(*) where ColName is null . This will help to eliminating UniqueIDs, Keys, and other similar data types. If all the rows are distinct, this would not be a good field for analysis. Same process for NULLs, if the percentage of nulls is greater than some threshold then you can eliminate those columns as well.
In order to automate it depending on your database, you could create a fairly simple stored procedure or function that loops through all the tables and columns and does a data type, count_distinct and a null percentage analysis on each field.
Once you've narrowed down list of columns, you can consider a .corr() function to run the analysis against all the remaining columns in something like a Python script.
If you wanted to keep everything in the database, Postgres also supports a corr() aggregate function, but you'll only be able to run this on 2 columns at a time, like this:
SELECT corr(column1,column2) FROM table;
so you'll need to build a procedure that evaluates multiple columns at once.
Thought about this tech challenges for some time. In general it’s AI solvable problem since there are easy features to extract such as unique values, clustering, distribution, etc.
And we want to bake this ability in https://columns.ai, obviously we haven’t gotten there yet, the first step we have done though is to collect all columns stats upon a data connection, identify columns that have similar range of unique values and generate a bunch of query templates for users to explore its dataset.
If interested, please take a look, as we keep advancing this part, it will become closer to an AI model to find relevant columns. Cheers!

Does a data warehouse need to satisfy 2NF or another normal form?

I'm investigating data warehouses. And I have an issue about star schemas.
It's in
Oracle® OLAP Application Developer's Guide
10g Release 1 (10.1)
3.2.1 Dimension Table: TIME_DIM
https://docs.oracle.com/cd/B13789_01/olap.101/b10333/global.htm#CHDCGABE
To represent the hierarchy MONTH -> QUARTER -> YEAR, we need some keys such as: YEAR_ID, QUARTER_ID. But there are some things that I do not understand:
1) Why do we need field YEAR_DSC & QUARTER_DSC? I think that we can look up these values from YEAR & QUARTER TABLE. And it breaks 2NF.
2) What is the normal form that a schema in data warehouse needs to satisfy? (1NF, 2NF, 3NF, or any.)
NFs (normal forms) don't matter for data warehouse base tables.
We normalize to reduce certain kinds of redundancy so that when we update a database we don't have to say the same thing in multiple places and so that we can't accidentally erroneously not say the same thing where it would need to be said in multiple places. That is not a problem in query results because we are not updating them. The same is true for a data warehouse's base tables. (Which are also just queries on its original database's base tables.)
Data warehouses are usually optimized for reading speed, and that usually means some denormalization compared to the original database to avoid recomputation at the expense of space. (Notice though that sometimes rereading something bigger can be slower than reading smaller parts and recomputing the big thing.) We probably don't want to drop normalized tables when moving to a data warehouse, because they answer simple queries and we don't want to slow down by recomputing them. Other than those tradeoffs, there's no reason not to denormalize. Some particular warehouse design methods might have their own rules about what parts should be denormalized what amounts.
(Whatever our original database design NF is chosen to be, we should always first normalize to 5NF then consciously denormalize. We don't need to normalize or know constraints to update or query a database.)
Read some textbook basics on why we normalize & why we use data warehouses.

Fact table design guidance for 100s of facts

I'm trying to create a datamart for the healthcare application. The facts in the datamart are basically going to be measurements and findings related to heart, and we have 100s of them. Starting from 1000 and can go to as big as 20000 per exam type.
I'm wondering what my design choices for the fact tables are:
Grain: 1 row per patient per exam type.
Some of the choices that I can think of -
1) A big wide fact table with 1000 or more columns.
2) EAV based design - A separate Measure dimension table. This foreign key will go into the fact table and the measure value will be in fact table. So the grain of the fact table will be changed to 1 row per patient per exam type per measurement.
3) Create smaller multiple fact tables per exam type per some other criteria like subgroup. But the end user is going to query across subgroups for that exam type and fact-fact join is not recommended.
4) any other ideas?
Any inputs would be appreciated.
1. A big wide fact table with 1000 or more columns.
One very wide fact table gives end-user maximum flexibility if queries are executed directly in the data warehouse. However some considerations should be taken into account, as you might hit some limits depending on a platform.
SQL Server 2014 limits are as per below:
Bytes per row 8,060. A row-overflow storage might be a solution, however it supports only few column types typically not related to fact nature, i.e. varchar, nvarchar, varbinary, sql_variant. Also not supported in In-Memory OLTP. https://technet.microsoft.com/en-us/library/ms186981(v=sql.105).aspx
Columns per non-wide table 1024. Wide-tables and sparse columns are solution as columns per wide table limit is 30,000. However the same Bytes per row limit applies. https://technet.microsoft.com/en-us/library/cc280604(v=sql.120).aspx
Columns per SELECT/INSERT/UPDATE statement 4,096
Non-clustered indexes per table 999
https://technet.microsoft.com/en-us/library/ms143432(v=sql.120).aspx
2. EAV based design - A separate Measure dimension table. This foreign key will go into the fact table and the measure value will be in fact table. So the grain of the fact table will be changed to 1 row per patient per exam type per measurement.
According to Kimball, EAV design is called Fact Normalization. It may make sense when a number of measurements is extremely lengthy, but sparsely populated for a given fact and no computations are made between facts.
Because facts are normalized therefore:
Extensibility is very easy, i.e. it's easy to add new measurements without the need to amend the data structure.
It's good to extract all measurements for one exam and present measurements as rows on the screen.
It's hard to extract/aggregate/make computation between several measurements (e.g. average HDL to CHOL ration) and present measurements/aggregates/computations as columns, i.e. requires complex WHERE/PIVOTING or multi-joins. SQL makes it difficult to make computations between facts in different rows.
If primary end-user platform is an OLAP cube then Fact Normalization makes sense. The cubes allows to make computation across any dimension.
Data importing could be an issue if data format is in a flat style CSV.
This questions is also discussed here Should I use EAV model?.
3) Create smaller multiple fact tables per exam type per some other criteria like subgroup. But the end user is going to query across subgroups for that exam type and fact-fact join is not recommended.
In some scenarios multiple smaller fact tables perfectly makes sense. One of the reason is if you hit some physical limits set by platform, e.g. Bytes per row.
The facts could be grouped either by subject area, e.g. measurement group/subgroup, or by frequency of usage. Each table could be placed on a separate file group and drive to maximize I/O.
Further, you could duplicate measurements across different fact tables to reduce the need of fact tables join, i.e. put one measurement in a specific measurement subgroup fact table and in frequently used measurement fact table.
However some considerations should be taken into account if there are some specific requirements for data loading. For example, if a record errors out in your ETL to one fact table, you might want to make sure that the corresponding records in the other fact tables are deleted and staged to your error table so you don't end up with any bogus information. This is especially true if end users have their own calculations in the front end tool.
If you use OLAP cubes then multiple fact tables actually becomes a source of a measure group to a specific fact table.
In terms of fact-to-fact join, you (BI application) should never issue SQL that joins two fact tables together across the fact table’s foreign keys. Instead, the technique of Drilling Across two fact tables should be used, where the answer sets from two or more fact tables are separately created, and the results sort-merged on the common row header attribute values to produce the correct result.
More on this topic: http://www.kimballgroup.com/2003/04/the-soul-of-the-data-warehouse-part-two-drilling-across/
4) any other ideas?
SQL XML or some kind NoSQL could be an option, but the same querying / aggregation / computation / presentation issues exist.

Designing a Data Warehouse/ Star Schema - Choosing facts

Consider a crowdfunding system whereby anyone in the world can invest in a project.
I have the normalized database design in place and now I am trying to create a data warehouse it (OLAP).
I have come up with the following:
This has been denormalized and I have chosen Investment as the fact table because I think the following examples could be useful business needs:
Look at investments by project type
Investments by time periods i.e. total amount of investments made per week etc.
Having done some reading (The Data Warehouse Toolkit: Ralph Kimball) I feel like my schema isn't quite right. The book says to declare the grain (in my case each Investment) and then add facts within the context of the declared grain.
Some facts I have included do not seem to match the grain: TotalNumberOfInvestors, TotalAmountInvestedInProject, PercentOfProjectTarget.
But I feel these could be useful as you could see what these amounts are at the time of that investment.
Do these facts seem appropriate? Finally, is the TotalNumberOfInvestors fact implicitly made with the reference to the Investor dimension?
I think "one row for each investment" is a good candidate grain.
The problem with your fact table design is that you include columns which should actually be calculations in your data-application ( olap cube ).
TotalNumberOfInvestors can be calculated by taking the distinct count of investors.
TotalAmountInvestedInProject should be removed from the fact table because it is actually a calculation with assumptions. Try grouping by project and then taking the sum of InvestmentAmount, which is a more natural approach.
PercentOfProjectTarget is calculated by taking the sum of FactInvestment.InvestmentAmount divided by the sum of DimProject.TargetAmount. A constraint for making this calculationwork is having at least a member of DimProject in your report.
Hope this helps,
Mark.
Either calculate these additional measures in a reporting tool or create a set of aggregated fact tables on top of the base one. They will be less granular and will reference only a subset of dimensions.
Projects seem to be a good candidate. It will be an accumulating snapshot fact table that you can also use to track projects' life cycle.

Would like to Understand 6NF with an Example

I have just read #PerformanceDBA's arguments re: 6NF and E-A-V. I am intrigued. I had previously been skeptical of 6NF as it was presented as "merely" sticking some timestamp columns on tables.
I have always worked with a data dictionary and do not need to be convinced to use one, or to generate SQL code. So I expect an answer that would require a dictionary (or catalog) that is used to generate code.
So I would like to know how 6NF would deal with an extremely simple example. A table of items, descriptions and prices. The prices change over time.
So anyway, what does the Items table look like when converted to 6NF? What is the "explosion of tables?" that happens here?
If the example does not work with a table this simple, feel free to add what is necessary to get the point across.
I actually started putting an answer together, but I ran into complications, because you (quite understandably) want a simple example. The problem is manifold.
First I don't have a good idea of your level of actual expertise re Relational Databases and 5NF; I don't have a starting point to take up and then discuss the specifics of 6NF,
Second, just like any of the other NFs, it is variegated. You can just barely step into it; you can implement 6NF for certan tables; you can go the full hog on every table, etc. Sure there is an explosion of tables, but then you Normalise that, and kill the explosion; that's an advanced or mature implementation of 6NF. No use providing the full or partial levels of 6NF, when you are asking for the simplest, most straight-forward example.
I trust you understand that some tables can be "in 5NF" while others are "in 6NF".
So I put one together for you. But even that needs explanation.
Now SQL barely supports 5NF, it does not support 6NF at all (I think dportas says the same thing in different words). Now I implement 6NF at a deep level, for performance reasons, simplified pivoting (of entire tables; any and all columns, not the silly PIVOT function in MS), columnar access, etc. For that you need a full catalogue, which is an extension to the SQL catalogue, to support the 6NF that SQL does not support, and maintain data Integrity and business Rules. So, you really do not want to implement 6NF for fun, you only do that if you have a need, because you have to implement a catalogue. (This is what the EAV crowd do not do, and this is why most EAV systems have data integrity problems. Most of them do not use the declarative Referential & Data Integrity that SQL does have.)
But most people who implement 6NF don't implement the deeper level, with a full catalogue. They have simpler needs, and thus implement a shallower level of 6NF. So, let's take that, to provide a simple example for you. Let's start with an ordinary Product table that is declared to be in 5NF (and let's not argue about what 5NF is). The company sells various different kinds of Products, half the columns are mandatory, and the other half are optional, meaning that, depending on the Product Type, certain columns may be Null. While they may have done a good job with the database, the Nulls are now a big problem: columns that should be Not Null for certain ProductTypes are Null, because the declaration states NULL, and their app code is only as good as the next guy's.
So they decide to go with 6NF to fix that problem, because the subtitle of 6NF states that it eliminates The Null Problem. Sixth Normal Form is the irreducible Normal Form, there will be no further NFs after this, because the data cannot be Normalised further. The rows have been Normalised to the utmost degree. The definition of 6NF is:
a table is in 6NF when the row contains the Primary Key, and at most one, attribute.
Notice that by that definition, millions of tables across the planet are already in 6NF, without having had that intent. Eg. typical Reference or Look-up tables, with just a PK and Description.
Right. Well, our friends look at their Product table, which has eight non-key attributes, so if they make the Product table 6NF, they will have eight sub-Product tables. Then there is the issue that some columns are Foreign Keys to other tables, and that leads to more complications. And they note the fact that SQL does not support what they are doing, and they have to build a small catalogue. Eight tables are correct, but not sensible. Their purpose was to get rid of Nulls, not to write a little subsytem around each table.
Simple 6NF Example
Readers who are unfamiliar with the Standard for Modelling Relational Databases may find IDEF1X Notation useful in order to interpret the symbols in the example.
So typically, the Product Table retains all the Mandatory columns, especially the FKs, and each Optional column, each Nullable column, is placed in a separate sub-Product table. That is the simplest form I have seen. Five tables instead of eight. In the Model, the four sub-Product tables are "in 6NF"; the main Product table is "in 5NF".
Now we really do not need every code segment that SELECTs from Product to have to figure out what columns it should construct, based on the ProductType, etc, so we supply a View, which essentially provides the 5NF "view" of the Product table cluster.
The next thing we need is the basic rudiments of an extension to the SQL catalog, so that we can ensure that the rules (data integrity) for the various ProductTypes are maintained in one place, in the database, and not dependent on app code. The simplest catalogue you can get away with. That is driven off ProductType, so ProductType now forms part of that Metadata. You can implement that simple structure without a catalogue, but I would not recommend it.
Update
It is important to note that I implement all Business Rules in the database. Otherwise it is not a database (the notion of implementing rules "in application code" is hilarious in the extreme, especially nowadays, when we have florists working as "developers"). Therefore all rules, etc are first and foremost implemented as SQL declarations, CHECK constraints, functions, etc. That preserves all Declarative Referential Integrity, and declarative Data Integrity. The extension to the SQL catalog covers the area that SQL does not have declarations for, and they are then implemented as SQL. Being a good data dictionary, it does much more. Eg. I do not write Views every time I change the tables or add or change columns or their characteristics, they are created directly from the catalog+extension using a simple code generator.
One more very important note. You cannot implement 6NF (or EAV properly, for that matter), without completing a full and faithful Normalisation exercise, to 5NF. The problem I see at every site is, they don't have a genuine 5NF state, they have a mish-mash of partial normalisation or no normalisation at all, but they are very attached to that. Creating either 6NF or EAV from that is a disaster. Creating EAV or 6NF from that without all business rules implemented in declarative SQL is a nuclear disaster, burning for years. You get what you pay for.
End update.
Finally, yes, there are at least four further levels of Normalisation (Normalisation is a Principle, not a mere reference to a Normal Form), that can be applied to that simple 6NF Product cluster, providing more control, less tables, etc. The deeper we go, the more extensive the catalogue. And higher levels of performance. When you are ready, just ask, I have already erected the models and posted details in other answers.
In a nutshell, 6NF means that every relation consists of a candidate key plus no more than one other (key or non-key) attribute. To take up your example, if an "item" is identified by a ProductCode and the other attributes are Description and Price then a 6NF schema would consist of two relations (* denotes the key in each):
ItemDesc {ProductCode*, Description}
ItemPrice {ProductCode*, Price}
This is potentially a very flexible approach because it minimises the dependencies. That's also its main disadvantage however, especially in a SQL database. SQL makes it hard or impossible to enforce many multi-table constraints. Using the above schema, in most cases it will not be possible to enforce a business rule that every product must always have a description AND a price. Similarly, you may not be able to enforce some compound keys that ought to apply (because their attributes could be split over multiple tables).
So in considering 6NF you have to weigh up which dependencies and integrity rules are important to you. In many cases you may find it more practical and useful to stick to 5NF and normalize no further than that.
I had previously been skeptical of 6NF
as it was presented as "merely"
sticking some timestamp columns on
tables.
I'm not quite sure where this apparent misconception comes from. Perhaps the fact that 6NF was introduced for the book "Temporal Data and The Relational Mode" by Date, Darwen and Lorentzos? Anyhow, I hope the other answers here have clarified that 6NF is not limited to temporal databases.
The point I wanted to make is, although 6NF is "academically respectable" and always achievable, it may not necessarily lead to the optimal design in every case (and not just when considering implementation using SQL either). Even the aforementioned discoverers and proponents of 6NF seem to agree e.g.
Chris Date: "For practical purposes, stick to 5NF (and 6NF)."
Hugh Darwen: "the 6NF decomposition around Date [not the person!] would be overkill... an optimal design for the soccer club is... 5-and-a-bit-NF!"
Hugh Darwen: "we are in 5NF but not in 6NF, and again 5NF is sufficient" (several similar examples).
Then again, I can also find evidence to the contrary:
Chris Date: "Darwen and I have both felt for some time that all base relvars should be in 6NF".
On a practical note, I recently extended the SQL schema of one of our products to add a minor feature. I adopted a 6NF to avoid nullable columns and ended up with six new tables where most (all?) of my colleagues would have used one table (or perhaps extended an existing table) with nullable columns. Despite me proving several 'helper' stored procs and a 'denormalized' VIEW with a INSTEAD OF triggers, every coder that has had to work with this feature at the SQL level has gone out of their way to curse me :)
These guys have it down: Anchor Modeling. Great academic papers on the subject, combined with practical examples. Their writings have finally pushed me over the edge to consider building a DW in 6nf on an upcoming project. The POC work I have done has validated (for me, at least) that the enormous benefits of 6nf don't outweigh the costs.

Resources