Semester 3 - Data Mining (Week 5-7)

2 minute read

Explain the categorization of visualization methods?

Data Visualization

  • Graphically represents data for clear communication.
  • Used for reporting, managing operations, tracking tasks, and discovering data relationships.

Visualization Techniques

image-center

A Pixel-oriented visualization

  • Pixel-Oriented: Uses color-shaded pixels to reflect data values and analyze correlations.

image-center

A Geometric projection visualization

  • Geometric Projection: Visualizes geometric transformations and projections of multidimensional data.

image-center

A Chernoff faces (Icon-based) visualization

  • Icon-Based: Uses icons, like Chernoff faces, to represent multidimensional data.
    • Chernoff Faces (Icon-Based)
      • Cartoon human faces representing up to 18 variables of multidimensional data.

image-center

A Tree map (Hierarchial) visualization

  • Hierarchical: Partitions dimensions into subsets visualized hierarchically. Methods include Info cube and screen filling method, Tree maps and Infocube.

Visualizing Complex Data and Relations

  • Techniques now include non-numeric data like text and social networks.
  • Tag clouds visualize statistics of user-generated tags.
  • Methods exist for visualizing relationships like social networks.

What is Data Quality?

  • Defined in terms of accuracy, completeness, consistency, timeliness, believability, and interpretability.
  • Quality is assessed based on the intended use of the data.

Factors Affecting Data Quality

  • Poorly designed data entry forms with many optional fields.
  • Human and deliberate errors.
  • Data decay and inconsistencies.
  • Instrumentation and system errors.
  • Inadequate data usage.

Explain about Data Preprocessing

  • Involves cleaning, integrating, reducing, and transforming data.
  • Cleaning fills in missing values, smooths noise, identifies outliers, and corrects inconsistencies.
  • Integration merges data from multiple sources.
  • Reduction minimizes data size while preserving information.
  • Transformation adjusts data for optimal mining.

Summarize Data Cleaning

  • Involves filling missing values, smoothing noise, identifying outliers, and correcting inconsistencies.
  • Typically an iterative two-step process: discrepancy detection and data transformation.

What is Integration of Data?

  • Combines data from different sources into one place.
  • Involves sorting out differences in meaning, managing data about data, checking relationships, finding and handling duplicate data and conflicts.

Define Reduction and Data Transformation

  • Part of data mining process.
  • Aims to decrease data volume while maintaining similar analytical outcomes.
  • Simplifies data for easier understanding and interpretation.
  • Techniques include reducing dimensions, reducing number of data points, and data compression.

What is Euclidean Distance?

  • Measures straight line distance between two points in a space.
  • Derived from Pythagoras’ theorem, used in data mining and machine learning.
  • Calculated using square root of sum of squares of differences in each dimension.
  • In 2D space, Euclidean distance between points (x1, y1) and (x2, y2) is sqrt((x2-x1)² + (y2-y1)²).
  • In 3D space, it extends to sqrt((x2-x1)² + (y2-y1)² + (z2-z1)²).
  • For higher dimensions, it’s sqrt(Σ(xi-yi)²), summing over all dimensions.

Define Different Distance Measures

  • Euclidean Distance: Straight line distance between two points.
  • Manhattan Distance: Distance between points along orthogonal axes (grid-based).
  • Chebyshev Distance: Maximum absolute distance in one dimension.
  • Minkowski Distance: Generalized metric distance measure with power p.
  • Hamming Distance: Minimum substitutions to change one string into another.
  • Mahalanobis Distance: Distance between a point and a distribution.

Data Similarity and Dissimilarity

  • Data similarity measures the likeness between two data objects.
  • It is subjective and defined based on the context.
  • Typically represented as a distance, with smaller distances indicating higher similarity.
  • Dissimilarity refers to the unlikeness or differences between data objects.

Define Data Visualization

  • Graphical representation of information and data.
  • Utilizes various visual tools like charts, graphs, and infographics.
  • Facilitates clear and efficient communication of information.
  • Enables quick analysis and exploration of large data sets for decision-making.

Data Discretization

  • Converts continuous data to discrete form.
  • Improves data understandability and interpretability.
  • Enhances machine learning model performance.
  • Methods include binning, histogram analysis, decision trees, and clustering.

You May Also Enjoy