How do you normalize a decision matrix before applying the entropy method?

Conditions

• The matrix contains nonnegative numerical attribute values.
• The sum of each column must be nonzero.
• Normalization is applied independently to each criterion column.

Reasoning, step by step

1. Calculate the sum of all entries in the first column.
2. Divide each entry in that column by the calculated column sum.
3. Repeat this process for every remaining column in the matrix.
4. Verify that the sum of the normalized values in each column equals 1.

Example

In the mobile-phone example, the Price or Cost column has a sum of 1250. The first entry, 250, is divided by 1250 to yield a normalized value of 0.2000.

Common misconceptions

• Normalizing by dividing each entry by the sum of its row instead of its column.
• Assuming the normalized values will sum to 1 across the entire matrix rather than within each individual column.

Watch the explanation

Explore next

• What is the entropy formula used after normalization and what does h represent?
• Why does the video say the entropy method does not need pairwise comparison opinions like AHP?