1/41
Comprehensive vocabulary flashcards covering key terms, concepts, architectures, and algorithms from the ISP642 Business Intelligence and Data Mining lecture notes.
Business Intelligence (BI)
An umbrella term that combines architectures, tools, databases, analytical tools, applications and methodologies for collecting, storing and analysing available data about an organisation’s operations to support better decision-making.
Strategic Gap
The difference between an organisation’s current performance and its desired performance, guided by its mission, objectives, goals, and strategy.
Data Warehouse
A foundational component of BI architecture that stores consolidated source data collected from various different operational sources.
Business Performance Management (BPM)
A core component of BI architecture dedicated to monitoring, measuring, and analysing business performance against defined organisational targets.
Cube Analysis
A style of BI that provides multidimensional analysis by slicing and dicing stored data from different perspectives and angles.
Enterprise Reporting
A BI style that utilizes structured dashboards and scorecards to visually present and communicate overall business performance.
Data Mining
The nontrivial process of identifying valid, novel, potentially useful and ultimately understandable patterns in data stored in structured databases using statistical, mathematical, and artificial intelligence techniques.
Extract, Transform and Load (ETL)
The first foundational element of data mining, responsible for transferring raw transaction data from operational sources into the data warehouse system.
Market Basket Analysis
A straightforward association technique used to understand which products or services are commonly purchased together by customers.
Customer Churn
A data mining application designed to predict which existing customers are likely to discontinue service or leave the company for a competitor.
Trend Analysis
A data mining technique that identifies shifts in customer behaviour across distinct time periods, such as comparing typical habits this month versus last month.
CRISP-DM
Cross-Industry Standard Process for Data Mining; a six-step process model comprising Business Understanding, Data Understanding, Data Preparation, Modelling, Testing and Evaluation, and Deployment.
Data Consolidation
The initial stage of data preparation in CRISP-DM that focuses on collecting, selecting, and integrating relevant data sources.
Data Cleaning
The stage of data preparation that involves imputing missing values, reducing data noise, and eliminating inconsistencies.
Data Transformation
The data preparation phase in which data is normalised, discretised, aggregated, or transformed to construct new analytical attributes.
Data Reduction
The data preparation step aimed at reducing the total number of variables and cases, as well as balancing skewed datasets.
SEMMA
A five-step core data mining process developed to guide technical execution: Sample, Explore, Modify, Model, and Assess.
KDD (Knowledge Discovery in Database)
A seven-step process to extract knowledge using specified thresholds: Data Cleaning, Integration, Selection, Transformation, Data Mining, Pattern Evaluation, and Knowledge Presentation.
Supervised Learning
A learning method that utilizes labelled training data containing input features and predetermined target/output variables to model relationships for predictive tasks.
Unsupervised Learning
A learning method that processes descriptive, unlabelled data without any predefined target variable to detect natural groupings and hidden relationships.
Classification
A supervised learning task that predicts discrete class labels or group membership by evaluating similarities between input instances and historical examples.
Regression
A supervised predictive task that estimates continuous numerical values for a dependent variable using known or fixed values of independent variables.
Association
An unsupervised data mining technique used to extract interesting correlations, frequent patterns, and association or causal structures among sets of items in repositories.
Link Analysis
An association technique that identifies connections and relationships among items, utilising methods such as Expectation Maximisation (EM), Apriori, and Graph-Based Matching.
Sequence Analysis
A technique that discovers time-oriented events to anticipate behavioural patterns, sequential transitions, and trends over time.
Apriori Algorithm
A classic algorithm for association rule learning that operates on transaction databases to find subsets of items meeting a specified minimum support threshold.
Support Count
The exact number of transactions in a transaction database that contain a given itemset.
Graph-Based Matching
The process of finding structural correspondence between nodes and edges of two graphs under specific constraints to map similar substructures between them.
Decision Tree Analysis
An automatic rule discovery technique that creates a flowchart-like, branching sequence of decisions leading to a discrete class outcome.
Bayesian Classifier
A probabilistic classification algorithm that assumes independence among predictive attributes and computes class-membership probabilities for each group.
Support Vector Machine (SVM)
A supervised learning algorithm based on statistical theory that projects data into a higher dimension via nonlinear mapping to find an optimal linear separating hyperplane.
Artificial Neural Network (ANN)
An interconnected network of artificial neurons inspired by biological neural systems that estimates complex non-linear functions through connectionist computation.
Artificial Immune Networks (AIN)
Adaptive computational systems that draw on models and metaphors from the human immune system (such as clonal and negative selection) to solve complex problem domains.
Rough Sets
An extension of classical set theory that addresses imprecise and uncertain knowledge by applying indiscernibility relations through lower and upper approximations.
Lower Approximation
In rough set theory, the set of all data instances that definitely and surely belong to a specified target class or concept.
Upper Approximation
In rough set theory, the set of all data instances that possibly belong to a specified target class or concept.
Genetic Algorithms (GA)
Directed search and optimisation techniques inspired by natural evolution that iteratively evolve a population of solutions using fitness evaluation, selection, crossover, and mutation.
Confusion Matrix
The primary analytical evaluation table used in classification problems to assess how correctly and reliably a predictive model assigns instances to classes.
Linear Regression
A statistical method for predicting the average value of a continuous dependent variable from one or more independent variables using a linear mathematical equation.
Regression Trees
A predictive modeling technique that applies recursive partitioning in a tree format, where each terminal leaf node corresponds to a local model tailored to that partitioned cell.
K-Means
An unsupervised clustering algorithm that partitions unlabeled data points into groups based on similarity to identify natural underlying clusters.
Self-Organising Map (SOM)
An unsupervised neural network technique used to organize and cluster data onto lower-dimensional grids, commonly applied in image segmentation and face detection.