Algorithm
## Algorithm

A description of how a process is executed.

×
Association analyses
## Association analyses

Analyses that study links between different things, such as market basket analysis.

×
Bayesian methods
## Bayesian methods

A branch of statistics with a back-to-front approach compared with the classical school: the findings are known but the reality is unknown. Modelling is usually based on simulation.

×
Business analytics
## Business analytics

A branch of analytics specialising in business development, aiming to find fact-based models for the promotion of business.

×
Business intelligence
## Business intelligence

Utilisation of data in business.

×
Classification methods
## Classification methods

A modelling technique in which new observational units are divided into predetermined categories on the basis of a model. The model is taught using data. Examples include modelling a phase of life with a neural network or decision tree.

×
Coefficient of determination
## Coefficient of determination

A percentage indicating how well data fit a model, i.e., the percentage of the variation in the dependent variable that can be explained by the model.

×
Confidence interval
## Confidence interval

An interval based on the sample that with a great certainty (for example with 95%) includes the key figure (average) of the population. A confidence interval defines the limits between which 95% of the averages will settle if the research would be repeated endlessly.

×
Confounding variable
## Confounding variable

Variables that affect both the confounding variable and the independent variable. For example, sometimes, between drownings and the sale of ice cream there is a strong correlation where the confounding variable is air temperature.

×
Control group
## Control group

The part of the target group that is not the target of any actions; for example, a marketing message is not sent to the members of the control group. Helps to measure the efficiency of an action.

×
Correlation
## Correlation

A measurement that indicates whether two variables are interdependent.

×
Cross tabulation
## Cross tabulation

Studies the distribution of invariables and their correlations. The research question could be e.g., whether the number of visits to the store differ between men and women.

×
Cross-over study design
## Cross-over study design

A study design where the groups are contradictorily exposed to same processes and the differences are compared in the end.

×
Customer analytics
## Customer analytics

Modelling customer behaviour and using the data to make business decisions.

×
Data mining
## Data mining

A method that extracts the essential from a large volume of data.

×
Data science
## Data science

Processing and analysis of large volumes of data.

×
Data synchronisation
## Data synchronisation

Harmonisation of separate sources of data.

×
Decision-making window
## Decision-making window

An easy-to-use and informative user interface for making business decisions using analytics.

×
Dependent variable (response variable)
## Dependent variable (response variable)

A variable that is modelled using other variables.

×
Descriptive analytics
## Descriptive analytics

Compression of data through visualisation and key figures.

×
Experimental design
## Experimental design

A plan that is drawn up before conducting an analysis. This preparatory stage involves planning how the data for an analysis should be collected from the test and control groups, for example.

×
Frequency distribution
## Frequency distribution

Distribution is formed with the number of occurring values i.e. frequencies. Can be depicted with a histogram, for example.

×
Grouping methods (segmentation, clustering)
## Grouping methods (segmentation, clustering)

Division of units (customer, product, etc,) into groups; the groups are homogeneous but different from the other groups.

×
Hypothesis testing methods
## Hypothesis testing methods

Methods for testing whether two groups are dissimilar or whether a variable is a relevant explanatory variable in a model, for example.

×
Industrial internet (Internet of things)
## Industrial internet (Internet of things)

Collecting data from devices and machines.

×
Machine learning
## Machine learning

Constructing rules and models from large volumes of data; examples include neural networks and support vector machines.

×
Market basket analysis
## Market basket analysis

An analysis that describes what products coexist in a shopping basket. On the basis of connections between products, it is possible to plan cross-sales, shelf locations and advertising.

×
Missing data
## Missing data

Gaps in data that can be filled in by statistical imputation methods.

×
Moving average
## Moving average

An average computed over a specific time window. This smooths out the data and reveals trends.

×
Multivariate methods
## Multivariate methods

Statistical models that use a large number of variables. Examples include the compression of a large number of variables through a principal components analysis and detecting non-measurable latent variables through a factor analysis.

×
Neural network
## Neural network

Mathematical model that is based on learning. The correlations between variables are taught to the neural network from the observations data and future observations can be predicted based on this.

×
Noise
## Noise

Data that contain random values that are difficult to model.

×
Optimisation (mathematical)
## Optimisation (mathematical)

Finding the most advantageous solution with the available resources.

×
Outlier
## Outlier

An observation point in data that is too distant from the other observations.

×
Path analysis
## Path analysis

An analysis that studies the order of steps taken to reach the final result or destination. For example, what web page the visitor to a page accessed next.

×
Prediction model
## Prediction model

A statistical algorithm or formula that uses known data for modelling connections between variables. Can also be extended outside the data.???

×
Predictive analytics
## Predictive analytics

Predicting the future on the basis of data detected by modelling.

×
Prescriptive analytics
## Prescriptive analytics

Optimizing decisions in a way that leads to the desired result.

×
Regression analysis
## Regression analysis

A commonly used modelling technique, in which the variable to be explained is presented with the help of explanatory variables and coefficients, often linearly.

×
Root cause analysis
## Root cause analysis

An analysis that aims to identify the cause of a phenomenon. Often confused with correlation.

×
Sample
## Sample

If the entire target group cannot be collected or processed, the analysis is carried out using a representative sample.

×
Scoring
## Scoring

An artificial indicator, used for purposes such as placing customers in a specific order. Can be produced through modelling, for example.

×
Significance level
## Significance level

Describes the probability of a statistical test to incorrectly reject null hypothesis. Most often used significance levels are 1%, 5%, and 10%.

×
Simulation methods
## Simulation methods

A technique aiming to imitate data through means such as a known distribution curve. Used for solving complex problems with computer power.

×
Standard deviation
## Standard deviation

Describes the observation values’ average deviation from the expectation.

×
Survey
## Survey

A set of questions that the members of the target group answer.

×
Survival analysis
## Survival analysis

A field of analytics that usually models the time remaining, using specific methods.

×
Text analysis
## Text analysis

Analysis of text in data, such as the identification of frequently occurring words, structures, categories, opinions and nuances.

×
Time series analysis
## Time series analysis

A branch of analytics that includes a time function. Requires specific modelling methods. Can be used for predicting the future.

×
Variation coefficient
## Variation coefficient

A dispersion figure that is not bound to unit of measurement. Variation coefficient is presented as percentage that is defined as the quotient of standard deviation and average.

×
Web analytics
## Web analytics

Mining and analysis of internet data, such as analysing customer behaviour with path analyses. It is often possible to identify customers and link web data to enrich other data sources.

×