Glossary

AI model training is the process of teaching an artificial intelligence system to make decisions or predictions using data.

AI model versioning involves maintaining the different versions of an AI model.

Big data refers to extremely large or intricate data sets that ordinary software cannot handle. AI uses complex algorithms to analyze this data and uncover patterns and correlations that humans might not notice.

An API, or Application Programming Interface, is a set of rules and protocols that enable different software applications to communicate with one another effectively.

API authorization is the process of determining whether a user or program is allowed to use a certain API (Application Programming Interface).

API calls are requests sent from one software program to another using an API.

An API endpoint is a unique web address (URL) that enables a client application to interact with an API.

API input is the data provided by the client to an API.

API rate limitation is an approach to manage incoming and outgoing traffic to and from a network, which includes APIs.

API integration is the practice of merging several APIs from various platforms or services to develop new functionality or improve existing software capabilities.

An API key is a code that computer programs use when calling an API to identify the program, its developer, or its user to the website.

ARIMA, or AutoRegressive Integrated Moving Average, is a forecasting method used for time series data.

Additive models are statistical methods used in time series analysis to distinguish between trends, seasonal variations, and other effects.

Advanced analytics goes beyond basic business intelligence by using advanced techniques to forecast future trends and behaviors.

Advanced data processing offers ways to deal with vast and complicated datasets that exceed the capabilities of ordinary software.

Artificial intelligence (AI) is a part of computer science that enables computers and machines to imitate human intelligence and solve problems. It includes machine learning and deep learning, which are about creating AI algorithms inspired by how the human brain works.

Automated Machine Learning automates the application of machine learning techniques to real-world issues by automating operations such as model selection, preprocessing, parameter tweaking, and more

Automated schema detection is a feature of data management systems that automatically identifies the dataset structure.

The base URL is the primary portion of a website's address, with no additional pathways or query parameters.

Bayesian ridge regression is a statistical method that applies Bayesian statistics to ridge regression, which is used to analyze data with multiple variables.

The Bernoulli distribution is a probability distribution used for generating random experiments with two possible outcomes, for example, success and failure, or 1 and 0.

Big data processing is the handling and analysis of extremely big and complex datasets that traditional data processing software cannot handle effectively.

Bottom-up forecasting is a prediction strategy that starts at the most granular level, such as individual sales or units, and then aggregates them to give a more comprehensive forecast.

Builder mode in software applications is a user-friendly interface that allows users to construct or alter apps without having to write code.

Business analytics is the use of data analysis tools and methodologies to make business decisions.

Business intelligence (BI) is the use of data analysis in business to aid and improve decision-making.

Change points in time series data refer to moments or periods where the statistical properties of the data significantly change.

Changepoint prior scale is a parameter in time series forecasting models that determines the model's flexibility in recognizing trends.

A classification sub-model is a component within a larger predictive model that is responsible for classifying data into predefined categories or labels.

Cloud integration involves linking multiple cloud services and connecting them to on-premises systems.

Cloud computing is the delivery of various computing services over the internet.

Cloud-based data preparation is the process of cleaning, transforming, and enriching data using cloud services.

Clustering in data analysis is the process of categorizing data into clusters in which items within the same cluster are more similar than those in other clusters.

Code debugging is the process of finding and fixing errors in software code.

The coefficient of determination (R^2) is a statistical measure that indicates how much variation in a dependent variable can be explained by the independent variables in a regression model.

In AI, cohorts represent distinct groups of data points or instances with similar attributes or characteristics, often used to train and evaluate machine learning models.

Collaborative AI refers to artificial intelligence systems or AI agents working together, either among themselves or with humans, to complete tasks or solve problems.

Column mapping is the process of matching the columns of one dataset with the columns of another.

Column mapping thresholds establish a criterion for determining when two columns from different datasets can be considered equal or matched.

Column weighting is the process of assigning different weights or importance to individual columns in a dataset during analysis.

Computational journalism is a field of journalism in which reporters use computer science techniques to collect, organize, analyze, and present news data.

A confidence interval is a range of values likely to contain an unknown population parameter.

Data Mapping is the process of connecting or associating data pieces from various sources or formats.

Data Matching is the process of comparing two datasets to identify similarities or matches between them.

Data modeling is the creation of visual or mathematical representations that demonstrate how data is structured and interrelated within a system or environment.

Data Operations (DataOps) is a methodology or approach that focuses on increasing the efficiency and quality of data analytics and delivery operations.

Data science is an interdisciplinary subject that uses scientific methods, algorithms, and tools to extract insights, patterns, as well as data from big datasets.

A database host is a server that manages the hardware and software environment in which a database operates.

In data science, a dataframe (sometimes known as a 'df') is a two-dimensional data structure similar to a table.

Decision trees are like visual maps that show decisions and their possible results in a tree shape.

Dimensionality reduction is a technique commonly used in statistical learning and data analysis to reduce the number of random variables being evaluated.

Fivetran is a service that allows enterprises to automate the migration of data from numerous sources into a single database or data warehouse.

Fuzzy matching is a method for identifying similarities between data points that may not be exact matches.

The Gaussian distribution, often known as the normal distribution, is a statistical concept that represents a symmetrical bell-shaped curve.

Hierarchical time series are structured arrangements of time-related data organized in a hierarchy or nested style.

IP whitelisting is a security practice that limits access to networks or systems based on approved IP addresses.

Ikigai is a Generative AI platform that transforms how businesses handle internal and external structured data, particularly timestamped information, turning it into actionable insights.

Imputation is a statistical technique used to fill in missing values within a data set by estimating or replacing them with plausible substitutes.

Jupyter Notebook is a free web application that lets you create and share documents including live code, equations, visualizations, and written explanations.

Though Large Language Models (LLMs) and Large Graphical Models (LGMs) both fall into the Generative AI category, they are structurally very different and serve unique purposes. LLMs primarily work on text, using sophisticated algorithms and neural networks to process and generate language while LGMs are designed for tabular and time series data and find and evaluate relationships across multiple dimensions.

Lars Lasso, also known as Least Angle Regression with the L1 Penalty, is a machine-learning technique for feature selection and regularization.

Lasso regression, or Least Absolute Shrinkage and Selection Operator, is a linear regression technique that incorporates L1 regularization.

Linear Discriminant Analysis (LDA) is a technique used in machine learning and pattern recognition to reduce dimensionality and classify data.

Linear extrapolation is a method for estimating the value of anything outside the known range by extending a straight line from existing data points.

A match score is a number that indicates how similar or dissimilar two objects are. It is used in a variety of applications, such as checking for duplication in databases or detecting similarities in the documents.

Matrix completion is a machine learning and data analysis technique that fills in missing values in a data matrix.

Mean Absolute Error (MAE) is a measurement that determines how well a predictive model can predict the outcomes.

Mean Absolute Percentage Error is similar to MAE but it is reported in percentage.

Mean Squared Error (MSE) is a popular metric used for assessing the accuracy of predictive models.

Model fine-tuning is the process of modifying a machine learning model's internal parameters or configurations to improve its accuracy while dealing with tasks or datasets.

Modular Data Processing: Modular data processing involves breaking down large data handling tasks into smaller and manageable sections known as modules.

Multidimensional forecasting is a strategy for predicting multiple related variables or outcomes over a period of time.

Multivariate Singular Spectrum Analysis (MSA): Multivariate Singular Spectrum Analysis is a statistical approach for analyzing complicated datasets containing several variables over time.

Naive Bayes is a type of machine learning algorithm that helps computers classify things into categories.

Natural Language Processing (NLP) is a part of artificial intelligence (AI) that deals with how computers and humans communicate using languages like English.

The Pandas library is a free tool for Python programmers who work with data.

A Pareto boundary refers to the set of optimal solutions in a multi-objective optimization problem where no solution can be improved in one objective without sacrificing performance in another.

A port number is like an address used by computers on a network to talk to each other.

Postman is an API development tool that makes it easier to create, test, and debug APIs.

Random Forest is a machine learning algorithm used for classification and regression tasks.

Regression analysis is a statistical method to study relationships between independent variables and a dependent variable.

Ridge regression is a regression analysis technique that avoids creating predictions that are overly particular to the data, which might result in erroneous results when applied to new data.

Scenario analysis is a planning method that helps firms consider many future scenarios.

Schema mapping is like creating a map for data. It helps computers understand how different pieces of data fit together by showing which parts of one set of data match with parts of another set.

Scientific computing is about using computers to solve big problems in science and engineering.

Seasonal Autoregressive Integrated Moving Average (SARIMA) is a statistical method for analyzing and forecasting time series data.

Seasonality Mode is a mathematical model for understanding and analyzing seasonal trends in a dataset.

Seasonality prior scale is a Bayesian modeling parameter that defines the impact of seasonal trends in data analysis.

Sentiment analysis is the computational process of finding, extracting, and evaluating subjective information and views conveyed in text data.

Singular Value Thresholding is a mathematical technique used in signal processing and matrix theory.

Structured data is organized and formatted in a predictable manner, typically within databases or spreadsheets, allowing for easy storage, retrieval, and analysis.

Synthetic data is artificially generated information that mimics real-world data characteristics but does not originate from actual observations.

Time series data captures how a value or outcome varies numerically over time. Each data point is associated with a specific timestamp or time period, enabling the identification of trends, patterns, and seasonality

Time series forecasting predicts future values using past data in chronological order.

Top-Down Forecasting is a forecasting technique that starts with a broad forecast for an organization or product category, which is then broken down into detailed estimations.

AI transformers are deep learning models equipped with self-attention mechanisms that revolutionize natural language processing tasks by effectively capturing contextual dependencies and semantic relationships within text data.

Unstructured data refers to information that lacks a predefined data model or organization, making it challenging to analyze using traditional methods.

Variance threshold removes low-variance features from a dataset, which are less useful for machine learning models.

Weighted Mean Absolute Percentage Error (WMAPE) is used to calculate forecast accuracy, particularly when some data points are more relevant than others.

Zero-shot classification is a type of machine learning task where a model learns to recognize classes it hasn't seen before during training.

aiCast is a forecasting AI model based on patented Large Graphical Models (LGM).

aiMatch is a patented AI model from Ikigai Labs that is based on Large Grapical Models (LGM) and designed for complex data reconciliation use cases, such as fraud detection or financial reconciliation.

aiPlan is a scenario planning AI model based on Large Graphical Models (LGM) which can generate and evaluate up to 10*19 scenarios based on complex datasets.

eXpert-in-the-loop (Xitl) refers to a hybrid approach in artificial intelligence where human expertise is integrated into the machine learning process.