Data Preprocessing

Data Preprocessing

1. Why Preprocess the data?

   • Consistency
     • Reduce representation of data set.
     • Dimension Reduction Numerosity Reduction
   • Data Transformation
     • Normalization,data discretization and concept hierarchy generation are forms of Data Transformation.
   • Completeness
     • Integrating multiple databases,data cubes or files.
   • Accuracy
     • There are many possible reasons for inaccurate data. For example,the age of a person must be enter within 100.
   • Believability
     • Believability reflects how much the data are trusted by users.
   • Interpretability
     • Interpretability reflects how easy the data are understood to users.

2. Major Tasks in Data Preprocessing

   • Data Cleaning
     • To Clean the data by filling in missing values,smoothing noisy data,identifying or removing outliers, and resolving inconsistencies.
     • 1. Missing Values. 2. Noisy Data. 3. Data Cleaning as a Process.
     • Missing Values
       • 1. Ignore the tuple.
       • 2. Fill in the missing value manually.
       • 3. Use a global constant to fill in the missing value.
       • 5. Use the attribute mean or median for all samples belonging to the same class as the given tuple.
       • 6. Use the most probable value to fill in the missing value.
     • Noisy Data
       • 1. Binning  -> Smoothing by bin means.-> Smoothing by bin medians.-> Smoothing by bin boundaries.
       • 2. Regression -> Linear Regression
     • Data Cleaning as a Process
       • Rules Used: -> Unique Rule. -> Consecutive Rule. -> Null Rule.
       • Tools Used: -> Data Subscribing Tools. -> Data Auditing Tools. -> Data Migration Tools.
   • Data Integration
     • To integrate multiple databases,data cubes, or files.
     • Redundancy and Correlation Analysis
       • Redundancy can be detected by Correlation Analysis.
       • 1. Correlation Test for Nominal Data. 2. Correlation Coefficient for Numeric Data. 3. Covariance of Numeric Data.
     • Entity Identification Problem
     • Tuple Duplication
       • Duplication should also be detected at tuple level.
     • Data Value Conflict Detection and Resolution
       • Data Integration also involves the detection and resolution of  Data Value Conflicts.
   • Data Transformation and Data Discretization
     • Normalization, data discretization, and concept hierarchy generation are forms of Data Transformation.
     • Data Transformation Strategies
       • 1. Smoothing. 2. Attribute Construction. 3. Aggregation. 4. Normalization. 5. Discretization. 6. Concept Hierarchy Generation for nominal data.
     • Data Transformation by Normalization
       • -> Min-Max Normalization. -> Z- Score Normalization. -> Decimal Scaling.
     • Discretization by Binning
       • Binning is a top-down splitting technique based on a specified number of bins.
     • Discretization by Histograms
       • Histogram analysis is an unsupervised discretization technique because it does not use class information.
     • Discretization by Cluster,Decision Tree and Correlation Analyses
     • Concept Hierarchy Generation for Nominal Data
       • 1. Specification of a partial ordering of attributes explicitly at the schema level by users or experts.
       • 2. Specification of a portion of a hierarchy by explicit data grouping.
       • 3. Specification of set of attributes,but not of their partial ordering.
       • 4. Specification of only a partial set of attributes.
   • Data Reduction
     • To obtain the reduced representation of the data set.
     • Data Reduction Strategies
       • Dimensionality Reduction. Numerosity Reduction.
       • Data Compression. -> Lossless. -> Lossy.
     • Wavelet Transforms
       • Discrete Wavelet Transform.
     • Principal Components Analysis
       • Also called Karhunen - Loeve method.
     • Attribute Subset Selection
       • Reduces the data set size by removing irrelevant or redundant attributes.
       • -> Step Forward Selection. -> Step Backward Elimination. -> Combination of both. -> Decision Tree Induction.
     • Regression and Log - Linear Models
       • Linear Regression. Multiple Linear Regression. Log-Linear models.
     • Histograms
       • Use binning to approximate data distributions and are a popular form of data reduction.
       • -> Equal Width. -> Equal Frequency.
     • Clustering
       • Clustering consider data tuples as objects. Centroid distance is an alternative measure of cluster quality.
     • Sampling
       • Sampling can be used as a data reduction technique because it allows a large data set to be represented by a much smaller random data sample.

Imágenes y archivos del mapa