OLAP is a precise method belonging to Data mining.
Data mining is a broader term covering a number of methods (one of them is OLAP). OLAP can be run: in software, you can ask for OLAP in a similar way you ask for correlations, regressions, and other statistics. OLAP is in a software menu (or it is a function in a language) (some packages will use “cube” or “pivot” keywords to get the same output).
Instead, you cannot ask for a Data Mining execution because there is no output associated with “Data mining”, you have to invoke a method belonging to Data mining to get an output such as regression trees, association rules, clusters, support vector machines: each one has its execution in a software.
In a book on Data mining, you can find a number of methods, OLAP is one of them. For instance, in the book “Data mining: concepts and techniques / Jiawei Han, Micheline Kamber, Jian Pei. – 3rd ed. ISBN 978-0-12-381479-1 “, OLAP is just a section of the (Data mining) book.
OLAP summarizes/aggregates by the group while other data mining methods will try to find hidden patterns from non-aggregated data (= the detail is king). OLAP would approach managerial information as the summary is welcome, other data mining methods would approach research as the detail is required.
OLAP consists of three basic analytical operations: consolidation (roll-up), drill-down, and slicing and dicing in the online mode, whereas “Data Mining” is a broader topic to mine massive data to find meaningful patterns out of the data by using various analytical techniques or framework such as CRISP-DM, Statical Techniques, Predictive Modeling techniques, exploratory data analysis, etc.
Data Mining can be done both in the online and offline mode. Sometimes, in data mining, there would not be any predefined objective.
Imagine you need a report on the turnover of your customers, then the turnover by product, then another by period, ... the list could be continued at will.
The number of possibilities increases dramatically when you start combining the different aspects, for example, top 5 products per customer or customers' sales by country over the period.
So that you don't have to build hundreds of queries/reports/dashboards, everything is calculated in advance in an OLAP cube. A cube is like a large network in which key figures are calculated/aggregated for each combination (customer, product, period, ...). So you can query each combination interactively. This method of pre-aggregating all combinations into a cube is called OLAP. Data mining is classic business statistics, classification, correlation, simulation, ...
In many tools, an OLAP cube can be used as a data source for data mining. Since OLAP cubes have to be coherent, they also have high data quality. This is estimated in data mining.
In summary: With OLAP you look for connections and try to show cause and effect. With data mining, you examine data statistically.
This is so much fun because in my field (process mining) we often get confused with data mining. So we enter the discussion of what is a subset of what, which are specific techniques and which are general fields where these techniques fit, etc.
Back to the original question, and as other posters have said, OLAP is just one technique within Data Mining that can be used to easily analyze multidimensional data. The different data dimensions are arranged in cubes, which can be "sliced" and then "diced" from different viewpoints, thus allowing to quickly obtain summarized information. OLAP databases implement this approach and allow users to effectively perform this kind of analysis.
Depending on your needs, you might opt to use OLAP, but you may also need to apply clusterization, predictive or simply statistical approaches.
All of these are considered data mining techniques as well.
Predictive Analytics solutions utilize statistical algorithms and machine learning techniques to analyze historical data and make predictions about future events or outcomes.
OLAP is a precise method belonging to Data mining.
Data mining is a broader term covering a number of methods (one of them is OLAP). OLAP can be run: in software, you can ask for OLAP in a similar way you ask for correlations, regressions, and other statistics. OLAP is in a software menu (or it is a function in a language) (some packages will use “cube” or “pivot” keywords to get the same output).
Instead, you cannot ask for a Data Mining execution because there is no output associated with “Data mining”, you have to invoke a method belonging to Data mining to get an output such as regression trees, association rules, clusters, support vector machines: each one has its execution in a software.
In a book on Data mining, you can find a number of methods, OLAP is one of them. For instance, in the book “Data mining: concepts and techniques / Jiawei Han, Micheline Kamber, Jian Pei. – 3rd ed. ISBN 978-0-12-381479-1 “, OLAP is just a section of the (Data mining) book.
OLAP summarizes/aggregates by the group while other data mining methods will try to find hidden patterns from non-aggregated data (= the detail is king). OLAP would approach managerial information as the summary is welcome, other data mining methods would approach research as the detail is required.
Great question @Evgeny Belenky. Thank you @Xavier Suriol for the answer.
OLAP consists of three basic analytical operations: consolidation (roll-up), drill-down, and slicing and dicing in the online mode, whereas “Data Mining” is a broader topic to mine massive data to find meaningful patterns out of the data by using various analytical techniques or framework such as CRISP-DM, Statical Techniques, Predictive Modeling techniques, exploratory data analysis, etc.
Data Mining can be done both in the online and offline mode. Sometimes, in data mining, there would not be any predefined objective.
Imagine you need a report on the turnover of your customers, then the turnover by product, then another by period, ... the list could be continued at will.
The number of possibilities increases dramatically when you start combining the different aspects, for example, top 5 products per customer or customers' sales by country over the period.
So that you don't have to build hundreds of queries/reports/dashboards, everything is calculated in advance in an OLAP cube. A cube is like a large network in which key figures are calculated/aggregated for each combination (customer, product, period, ...). So you can query each combination interactively. This method of pre-aggregating all combinations into a cube is called OLAP. Data mining is classic business statistics, classification, correlation, simulation, ...
In many tools, an OLAP cube can be used as a data source for data mining. Since OLAP cubes have to be coherent, they also have high data quality. This is estimated in data mining.
In summary: With OLAP you look for connections and try to show cause and effect. With data mining, you examine data statistically.
This is so much fun because in my field (process mining) we often get confused with data mining. So we enter the discussion of what is a subset of what, which are specific techniques and which are general fields where these techniques fit, etc.
Back to the original question, and as other posters have said, OLAP is just one technique within Data Mining that can be used to easily analyze multidimensional data. The different data dimensions are arranged in cubes, which can be "sliced" and then "diced" from different viewpoints, thus allowing to quickly obtain summarized information. OLAP databases implement this approach and allow users to effectively perform this kind of analysis.
Depending on your needs, you might opt to use OLAP, but you may also need to apply clusterization, predictive or simply statistical approaches.
All of these are considered data mining techniques as well.