Time Series, Sequential Pattern Mining, and Network Analysis – Vocabulary

61 vocabulary flashcards covering fundamental terms from lectures on Time Series, Sequential Pattern Mining, and Network Analysis.

A sequence of data points ordered chronologically over a period of time.

Time Series

Technique that uses historical values and their patterns to predict future observations.

Time Series Forecasting

Modeling approach where current values depend on previous values (lags) of the same series; represented by parameter p in ARIMA.

Autoregression (AR)

Transformation that subtracts an observation from a prior one to remove trend/seasonality and achieve stationarity; parameter d in ARIMA.

Differencing

Time-series model that uses past forecast errors (lagged residuals); parameter q in ARIMA.

Moving Average (MA)

The long-term upward or downward movement in a series.

Trend (Time Series)

Long-term oscillations around the trend that usually last at least two years.

Cyclical Component

Regular, periodic fluctuations that repeat during the same period each year.

Seasonality

Time-series decomposition where observation equals T + S + C + I; seasonal amplitude independent of level.

Additive Model

Decomposition where observation equals T × S × C × I; seasonal amplitude varies with level.

Multiplicative Model

The unpredictable residual part of a time series after other components are modeled.

Irregular / Noise Component

Series with constant mean, variance, and autocorrelation over time.

Stationary Series

Plotting moving averages or variances to visually assess stationarity.

Rolling Statistics

Statistical test whose null hypothesis states the series is non-stationary (has a unit root).

Dickey-Fuller Test

Extension of the Dickey-Fuller test that includes lagged difference terms to detect a unit root.

Augmented Dickey-Fuller (ADF) Test

Kwiatkowski-Phillips-Schmidt-Shin test where the null hypothesis is trend stationarity.

KPSS Test

Auto-Regressive Integrated Moving Average model combining AR, differencing, and MA terms for forecasting.

ARIMA

Average of the absolute forecast errors.

Mean Absolute Error (MAE)

Average of the squared forecast errors, giving more weight to large errors.

Mean Squared Error (MSE)

Square root of MSE, expressing error in original units.

Root Mean Squared Error (RMSE)

Frequency at which observations in a time series are collected.

Sampling Rate

Data-mining task that discovers frequently occurring ordered events or subsequences in a set of sequences.

Sequential Pattern Mining (SPM)

Ordered list of itemsets or symbols in SPM.

Sequence

Single symbol within a sequence.

Item

Unordered set of distinct items occurring together at one time-stamp.

Itemset

Sequence derived by deleting zero or more items from another sequence without altering order.

Subsequence

Subsequence whose support meets or exceeds the minimum support threshold.

Sequential Pattern

Number of sequences in the database that contain a given sequence.

Support (SPM)

User-defined cut-off that a sequence’s support must reach to be considered frequent.

Minimum Support Threshold (minsup)

Principle stating that if a sequence is not frequent, none of its super-sequences can be frequent.

Apriori Property

Apriori-based algorithm that mines sequential patterns using a horizontal data format.

GSP (Generalized Sequential Pattern)

Apriori-based algorithm that mines sequential patterns using a vertical data format.

SPADE

Pattern-growth algorithm for sequential pattern mining that reduces candidate generation.

FreeSpan

Pattern-growth algorithm that projects databases based on sequence prefixes to find frequent patterns.

PrefixSpan

SPM type applied to data with limited alphabets, such as DNA sequences.

String Mining

SPM type focusing on ordered sets of itemsets, common in marketing and sales.

Itemset Mining (Sequential Context)

Implication X → Y indicating that if items in X occur, they are followed by items in Y.

Sequential Rule

Support of the rule divided by support of X; estimates the conditional probability P(Y|X).

Confidence (Sequential Rule)

Association measure where Lift > 1 indicates Y is more likely after X than under independence.

Lift

Dependence measure; values greater than 1 show strong dependence of X on Y.

Conviction

Association measure ranging from −1 (negative) to +1 (positive); near 0 suggests no association.

Zhang’s Metric

Collection of nodes connected by edges representing relationships.

Network (Graph)

Fundamental unit in a graph representing an entity.

Node (Vertex)

Connection between two nodes; may be directed or undirected.

Edge (Link)

Number of edges incident to a node.

Degree

Overall structural arrangement of connections in a graph.

Network Topology

Topology where every node has the same degree.

Regular Network

Graph whose edges are formed randomly between nodes.

Random Network

Network whose degree distribution follows a power law with a few highly connected hubs.

Scale-free Network

Group of nodes more densely connected internally than with the rest of the network.

Cluster / Community

Algorithms and techniques used to identify clusters within a graph.

Community Detection

Centrality metric equal to the number of connections a node has.

Degree Centrality

Metric counting the number of shortest paths that pass through a node, indicating its bridging role.

Betweenness Centrality

Centrality measure that rewards connections to other high-scoring nodes.

Eigenvector Centrality

Measure of how near a node is to all others based on shortest path lengths.

Closeness Centrality

Heuristic algorithm that maximizes modularity for community detection in large networks.

Louvain Method

Community detection algorithm that iteratively removes edges with highest betweenness centrality.

Girvan-Newman Method

Network growth model producing scale-free graphs via preferential attachment.

Barabási–Albert Model

Process of simulating or describing how networks form and evolve over time.

Network Modeling

Study of how nodes, edges, and network properties change over time.

Network Dynamics

Methodology for visualizing and analyzing relationships among entities in a social context.

Social Network Analysis (SNA)