The water supply association analysis method in Shenzhen based on kmeans clustering discretization and apriori...

Area of Science:

• Data Mining and Machine Learning
• Water Resource Management
• Supply Chain Analytics

Background:

• Water supply association analysis is crucial for understanding supply fluctuations but current methods using continuous data are ineffective.
• Monotone relationships and multicollinearity in continuous data distort analysis, leading to inaccurate results and hindering understanding of feature-water supply associations.
• Existing methods fail to provide clear association rules and value intervals, limiting effective water supply dispatching strategies.

Purpose of the Study:

• To develop an effective association analysis method for water supply data that overcomes the limitations of continuous data analysis.
• To identify valid strong association rules and value intervals between features and water supply.
• To provide better support for water supply dispatching by understanding causes and intervals of water supply fluctuation.

Main Methods:

• A data mining approach coupling K-means clustering discretization and the Apriori algorithm was proposed.
• K-means clustering was used for data discretization to generate one-hot encoding compatible with the Apriori algorithm.
• Discretization using K-means mitigates the influence of monotone relationships and multicollinearity, and all generated rules undergo validation.

Main Results:

• The proposed method effectively identifies valid strong association rules between features and water supply.
• The analysis reveals whether the association relationship is direct or indirect, offering deeper insights.
• The method successfully determines feature value intervals, association degrees, and rule confidence probabilities.

Conclusions:

• The coupled K-means and Apriori method is an effective approach for water supply association analysis.
• This technique provides valuable insights into feature-water supply relationships, including rules, intervals, and association strength.
• The findings support improved water supply fluctuation attribution analysis and enhance water supply dispatching decisions.