Data Preprocessing in Data Mining

Data Preprocessing

• Aggregation: combining two or more attributes (or objects) into a single attribute (or object)
  • Purpose: data reduction, change of scale, and more stable data
• Sampling: main technique for data reduction, used for preliminary investigation and final data analysis
  • Key principle: using a sample will work almost as well as using the entire data set, if the sample is representative
• Discretization and Binarization: transforming continuous data into discrete or binary form
• Attribute Transformation: changing the scale or format of data
• Dimensionality Reduction: reducing the number of attributes or features in the data
• Feature subset selection: selecting a subset of the most relevant features
• Feature creation: creating new attributes that can capture the important information in a data set more efficiently

Aggregation

• Purpose: data reduction, change of scale, and more stable data
• Examples: aggregating cities into regions, states, or countries, days into weeks, months, or years
• Effect: aggregated data tends to have less variability

Sampling

• Types: simple random sampling, sampling without replacement, sampling with replacement
• Sample size: affects the representativeness of the sample
• Importance: sampling is used to reduce the amount of data and make data analysis more efficient

Dimensionality Reduction

• Purpose: avoid curse of dimensionality, reduce data size, and enable visualization
• Techniques: Principal Components Analysis (PCA), Singular Value Decomposition, and others
• Effect: reduces the number of attributes or features in the data

Principal Components Analysis (PCA)

• Goal: find a projection that captures the largest amount of variation in data
• Result: reduces the dimensionality of the data while retaining most of the information

Feature Subset Selection

• Purpose: select a subset of the most relevant features
• Redundant features: duplicate much or all of the information contained in one or more other attributes
• Irrelevant features: contain no information that is useful for the data mining task
• Techniques: many techniques developed, especially for classification

Feature Creation

• Purpose: create new attributes that can capture the important information in a data set more efficiently
• Methodologies: feature extraction, feature construction, and mapping data to new space
• Examples: extracting edges from images, dividing mass by volume to get density, and Fourier and wavelet analysis

Data Exploration

• Statistical measurements: mean, median, mode, quantiles, quartile, percentile, frequency
• Visualization: x-y scatter, histogram, violin, heatmap, networks, hierarchical plots, spatio-temporal, contour, surface vector field plots, and Chernoff Faces
• Multivariate data: OLAP, DataCubes