A frequency distribution shows a frequency count and characteristics
for the distinct values in a column. When you run a column analysis job, a
frequency distribution is generated for each column that you analyze.
You can use a frequency distribution to learn about the content of your
data and make decisions about how your data is structured. For example, when
you want to learn about how your data is categorized, you can examine the
data classes that were inferred during analysis. If you do not agree with
the inferences, you can select another data class when you review the frequency
distribution.
A frequency distribution is also the foundation on which you run other
analyses when you profile data. For example, during a primary key analysis
job, a frequency distribution for a column is evaluated for unique values.
If a column is unique, it might be inferred as a candidate for the primary
key.
When a frequency distribution is created during analysis, statistics about
the column are generated. The statistics include information such as the length
of the characters in a column, nulls, and data properties.
A frequency distribution lists the characteristics of the columns in your
data:
- Inferences about the data type of each value in the column
- A data type describes the structural format of data in a column. For example,
columns that contain numeric data are type N (numeric), and columns that contain
alphabetic data are type A (alphabetic).
- Inferences about the data class of the column
- A class is a variable that categorizes a column according to how the data
in the column is used. For example, if a column contains data such as 10/04/07,
the Date class is assigned to the column because 10/04/07 is an expression
for a date.
- The length calculation for each value in the column
- Shows the total length of the values in a column.
- The number of nulls in the column
- Shows the total number of empty values in a column.
- The number of valid and invalid values in the column
- Shows whether a value has been inferred as valid or invalid.
- The format expression for each value in the column
- A format expression is a pattern that describes the type of data in a
column according to the type of the column. For example, if a column is of
the class Date, a format expression such as YYMMDD might be assigned to the
column (year, month, and day.)
The analysis also develops statistics about the columns in
your data:
- The cardinality count of the column
- A cardinality count is the total number of distinct values in a column.
- The cardinality percentage of the column
- A cardinality percentage is a calculation of the total number of distinct
values in a column divided by the total number of values in the same column.
- Minimum, maximum, and average length of values in the column
- Shows the average length of values and then calculates the minimum and
maximum lengths of values in the same column.
- Distribution of inferred data types
- Shows a distribution of all the data types that were inferred during analysis.
- Distribution of inferred format expressions
- Shows a distribution of all of the format expressions that were inferred
during analysis.
IBM InfoSphere Information Analyzer, Version 8.5
This topic is also in the IBM InfoSphere Information Analyzer User's Guide.
