A missing value is information that should exist for an entity or observation but is absent. In graph analysis, missing properties can describe incomplete records or undefined relationships between nodes. section 10.4.1 section 4.2
Analyze missing values by:
- Counting all records with count(*) and comparing them with non-null property values returned by count(property). section 4.2
- Reporting missingness as a percentage, casting an integer to float before division in Neo4j. section 4.2
- Choosing a replacement that reflects the meaning of the missing value rather than applying one universal default. section 10.4.1
User registration dates
Among 3,594 User nodes, 3,518 have a non-null registeredAt property, leaving 76 missing registration dates; 97.88% are populated. The createdAt property for Tweet nodes has no missing values. section 4.2
Listing 10.10 Filling in the missing values
data['networkDistance'].fillna(4, inplace=True) data['adamicAdar'].fillna(8.097444, inplace=True)
Fill undefined network distances with 4 and missing Adamic-Adar values with 8.097444 in a pandas DataFrame. section 10.4.1
