1

Topic: Network for data analysis in tables, probably?

All greetings;
There is a ton of tables and it is necessary to analyze the data on adequacy.
The totals at groupings on various indexes. A problem that is a lot of indexes.
The question - whether can a network or still what infernal mechanism, it to analyze and reveal deviations?
Pivot tables in  we are able to draw, but  it is a lot of indexes. It is heavy to person to supervise all it on adequacy.
Thanks.

2

Re: Network for data analysis in tables, probably?

Such network that understood that such adequacy, still is not present)

3

Re: Network for data analysis in tables, probably?

And what to you hinders to write stored procedure?

4

Re: Network for data analysis in tables, probably?

There sense in checking the data. Than the previous value differs from the leaking.
If distinction, say, is more 5 % it is necessary to show it to the user.
any fool can write, only as you represent it to yourselves? At me about 200 indexes - for everyone it is necessary to do the cursor? It .
Any tool that it easier to do is necessary.

5

Re: Network for data analysis in tables, probably?

User2155;
Result an adequate example of your tables, indexes, groupings and checks differently any distinct council you do not receive.

6

Re: Network for data analysis in tables, probably?

The grandfather;
Cube with dimensionalities:
- Time (months);
- Groups of products;
- Sales.
If the goods And two years were on sale in number of 200 pieces, and in September 2017 sold 112 most likely somewhere in basis an error and sales stupidly were not counted.
If at the company always was about 300 clients, and in their September, 2017 became 98, it most likely an error in calculations.
And it is a lot of such indexes. I should not do business an analytics - for this purpose other people are, I should check accuracy of calculations.

7

Re: Network for data analysis in tables, probably?

The table dataful is accessible to a cube.
I.e. it is possible to address directly in SQL or in SSAS a cube (as it is more convenient).

8

Re: Network for data analysis in tables, probably?

User2155;
Data Mining?
https://docs.microsoft.com/en-us/sql/an … ining-ssas
https://docs.microsoft.com/en-us/sql/an … s-services

9

Re: Network for data analysis in tables, probably?

User2155 wrote:

Cuba with dimensionalities:
- Time (months);
- Groups of products;
- Sales.
If the goods And two years were on sale in number of 200 pieces, and in September 2017 sold 112 most likely somewhere in basis an error and sales stupidly were not counted.
If at the company always was about 300 clients, and in their September, 2017 became 98, it most likely an error in calculations.

date-kvoliti not always do by "carpet bombing" checking all indexes.
But sometimes, yes all is necessary.
Time at you the data already cubed (a cube I imply SSAS?) similar checks are written on MDX simply enough.
You need to write once as a matter of fact and then many times to use (adding of checks and their modification it is not considered).
Strongly I suspect that checks for the different companies and products at you in the majority are same
(Comparing of a measure of the current period with value in the past on measurements the periods \companies \goods).
Therefore you most likely not 200 + indexes, and have less.
- It is possible to fist eggs and to write hands.
- It is possible to write a script on biml which on meta data  to you the code.

10

Re: Network for data analysis in tables, probably?

Where I will dig:
1) Data mining.
I will look at video, can works. At first sight it not absolutely that is necessary, but I only 40 minutes , can further will more interesting.
2) Command LAG on MS SQL.
- We create function which as parameter accepts the table (User Defined Type).
In the table only 2 columns: date and number. I.e. the data grouped in days, and in my case it always the end of month.
- Function calculates an average gain/reduction for a year;
- If for the last date value strongly deviates an average - a signal to the user;
While so. All thanks.