How to select a row for update values, Using CTEs or suitable method

Question

HI All,
I have a table as claim_data as below.
claim_id   
amount         change_monthkey
         created_monthkey
         ord
54511      
300            201304         
         201304          
          1
54511      
0              201305         
         201304          
          2
120301     
250            201502         
         201502          
          1
120624     
150            201502         
         201502          
          1
120624     
0              201503    
              201502          
          2
I want to isolate rows which appear in the table and do a update using below query ( This query is already in a procedure, so I just looking for a suitable way to modify it to support my task )
In the above e.g.  ONLY row containing claim_id = 120301 needs to update (amount  as updated_amount)
I use following query, but it update other rows as well, I want a suitable query only to capture rows of nature claim_id = 120301 and do this update? Is there any functions that I could use to facilitate my task??
select
     a.claim_id
    , a.Created_MonthKey
    , a.Change_MonthKey
    , a.amount - ISNULL(b.amount,0.00) as amount_movement
    ,a.amount  as updated_amount   --- >> I need help here??
    FROM claim_data a
LEFT JOIN claim_data b
    ON a.claim_id = b.claim_id
    AND b.Ord = a.Ord - 1
Thanks
Mira

Answer 1

Please follow basic Netiquette and post the DDL we need to answer this. Follow industry and ANSI/ISO standards in your datC1. You should follow ISO-11179 rules for naming data elements. You do not know this IT standard.
You should follow ISO-8601 rules for displaying temporal datC1. We need to know the data types, keys and constraints on the table. Avoid dialect in favor of ANSI/ISO Standard SQL. 
And you need to read and download the PDF for: 
https://www.simple-talk.com/books/sql-books/119-sql-code-smells/
>> I have a table AS claim_data AS below. << 
Thanks to your lack of netiquette, we have no name, no DDL, no keys and have to re-type the raw data into SQL, guess at everything! Does your boss treat you with the same contempt? 
Think about how silly “_data” is AS an attribute property! What tables do you have without an data in them? You had to give this meta-data in 1970's operating systems, FORTRAN compilers, etc. but we do not do this today. I see you use “A”, “B”, etc for table
aliases This is the name of the disk or tape drive in a 1970 computer! In SQL we use helpful, meaningful table aliases so code is easier to maintain. I see you also use the old punch card trick of putting a comma at the front of each line of code. 
The column named “ord” is a mathematical fiction for ordinal sets. I think you meant to use it as a non-relational sequential number to let you keep writing 1970's COBOL is T-SQL. We have not used the old Sybase ISNULL() since we got COALESCE() in SQL-92. They
are actually different. 
First, let's post what you should have posted if you followed forum rules: 
CREATE TABLE Claims
(claim_id CHAR(5) NOT NULL,
 claim_amount DECIMAL (8,2) NOT NULL
 CHECK (claim_amount >= 0.00),
 change_month_name CHAR(10) NOT NULL
 REFERENCES Month_Period(month_name),
 creation_month_name CHAR(10) NOT NULL
 REFERENCES Month_Period(month_name),
 ord INTEGER NOT NULL, -- NO!! Awful design! 
 PRIMARY KEY (claim_id, ord)); --required by definition
Since SQL is a database language, we prefer to do look ups and not calculations. They can be optimized while temporal math messes up optimization. A useful idiom is a report period calendar that everyone uses so there is no way to get disagreements in the DML.
The report period table gives a name to a range of dates that is common to the entire enterprise. 
CREATE TABLE Month_Periods
(month_name CHAR(10) NOT NULL PRIMARY KEY
 CHECK (month_name LIKE '[12][0-9][0-9][0-9]-[01][0-9]-00'),
 month_start_date DATE NOT NULL,
 month_end_date DATE NOT NULL,
 CONSTRAINT date_ordering
 CHECK (month_start_date <= month_end_date),
etc);
These report periods can overlap or have gaps. I like the MySQL convention of using double zeroes for months and years, That is 'yyyy-mm-00' for a month within a year and 'yyyy-00-00' for the whole year. The advantages are that it will sort with the ISO-8601
data format required by Standard SQL and it is language independent. It has been proposed for the Standard, but is not yet part of it. The regular expression pattern for validation is simple. 
INSERT INTO Claims  -- a mess! 
VALUES
('054511', 300.00, '2013-04-00', '2013-04-00', 1)
('054511', 0.00, '2013-05-00', '2013-04-00', 2),
('120301', 250.00, '2015-02-00', '2015-02-00', 1),
('120624', 150.00, '2015-02-00', '2015-02-00', 1),
('120624', 0.00, '2015-03-00', '2015-02-00', 2);
Now, throw this mess out. It looks like your repeat the claim creation date over and over and over. It looks like you crammed a claim creation and a claim history together in one table. 
>> I want to isolate rows which appear in the table and do a update using below query (This query is already in a procedure, so I just looking for a suitable way to modify it to support my task). 
In the above e.g. ONLY row containing claim_id = '120301' needs to update (amount AS updated_amount)
>> I use following query, but it update other rows as well, I want a suitable query only to capture rows of nature claim_id = '120301' and do this update? <<
No. SQL is set oriented so an update applies to the whole table. You can filter out rows in a WHERE clause. 
CREATE TABLE Claims
(claim_id CHAR(5) NOT NULL PRIMARY KEY,
 claim_amount DECIMAL (8,2) NOT NULL
 CHECK (claim_amount >= 0.00),
claim_creation_month_name CHAR(10) NOT NULL
 REFERENCES Month_Period(month_name));
See how this works? 
INSERT INTO Claims  -- a mess! 
VALUES
('054511', 300.00, '2013-04-00'),
('120301', 250.00, '2015-02-00'),
('120624', 150.00, '2015-02-00');
Normalize the table: 
CREATE TABLE Claim_Changes
(claim_id CHAR(5) NOT NULL 
  REFERENCES Claims(claim_id)
  ON DELETE CASCADE,
 change_month_name CHAR(10) NOT NULL
  REFERENCES Month_Period(month_name),
 PRIMARY KEY (claim_id, change_month_name),
 change_amount DECIMAL (8,2) NOT NULL
 CHECK (change_amount >= 0.00)
INSERT INTO Claims_Changes
VALUES
('054511', 0.00, '2013-05-00'),
('120624', 0.00, '2015-03-00');
Now you just add another change row to the history. There is no update. If you want to the delta and so forth, use LAG() and other window functions. 
--CELKO-- Books in Celko Series for Morgan-Kaufmann Publishing: Analytics and OLAP in SQL / Data and Databases: Concepts in Practice Data / Measurements and Standards in SQL SQL for Smarties / SQL Programming Style / SQL Puzzles and Answers / Thinking
in Sets / Trees and Hierarchies in SQL