Improving Table Designs Using Normalization.

Solution Preview

Let us start with a single table with all columns in it. Let's remember that, in all tables candidate keys (also known as Primary Keys) are in BOLD

Let's call the below table OrderInfo table:

Table OrderInfo:

Order_id
Order_date
Customer_id
Customer_name
Customer_address
Customer_city
Customer_state
Item_id
Item_description
Item_qty
Item_price
Item_total_price
Order_total_price

The following FDs (functional dependencies) hold on the above table:

Order_Id -> Customer_id, Order_date, Order_total_price
Item_Id, Order_Id -> Item_Qty, Item_total_price, Item_description, Item_Price
Customer_id -> Customer_Name, Customer_Address, Customer_City, Customer_State
Item_Id -> Item_description, Item_price

Primary key of the above table is a composite candidate key Order_id and Item_id

Clearly, this table is in 1 NF, but it is not in 2NF, so obviously not in 3 NF.

The table is not in 2 NF because we have FDs with part of candidate key Order_id and Item_id and also Customer_id

So, we have to separate the fields according to the FDs and create more tables.

Hence, at first we take the FD

Customer_id -> Customer_Name, Customer_Address, Customer_City, Customer_State

Then, we need to put the below fields in Customer table

Customer_id
Customer_name
Customer_address
Customer_city
Customer_state

In a Customer ...