WO2003012713A1 - Data visualisation system and method - Google Patents

Data visualisation system and method Download PDF

Info

Publication number
WO2003012713A1
WO2003012713A1 PCT/NZ2002/000138 NZ0200138W WO03012713A1 WO 2003012713 A1 WO2003012713 A1 WO 2003012713A1 NZ 0200138 W NZ0200138 W NZ 0200138W WO 03012713 A1 WO03012713 A1 WO 03012713A1
Authority
WO
WIPO (PCT)
Prior art keywords
time
data
values
vector
vectors
Prior art date
Application number
PCT/NZ2002/000138
Other languages
French (fr)
Inventor
Andrew John Cardno
Nicholas John Mulgan
Richard Thomas Beard
Original Assignee
Compudigm International Limited
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Compudigm International Limited filed Critical Compudigm International Limited
Publication of WO2003012713A1 publication Critical patent/WO2003012713A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2137Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on criteria of topology preservation, e.g. multidimensional scaling or self-organising maps

Definitions

  • the invention relates to a data visualisation system and method, particularly but not solely designed to identify patterns in time-variant data, for example stock exchange trading data.
  • the invention is particularly suitable in identifying patterns in time- variant multivariate data.
  • the invention comprises a data visualisation system comprising time-variant data maintained in computer memory; a time series analysis component configured to create one or more vectors from the time-variant data; a self- organising map component configured to generate and display a two-dimensional representation including one or more vector representations; and a contour generator configured to generate and display one or more contour lines around each vector representation.
  • a method of data visualisation comprising the steps of maintaining time-variant data in computer memory; creating one or more vectors from the time-variant data; generating and displaying a two-dimensional representation including one or more vector representations; and generating and displaying one or more contour lines around each vector representation.
  • Figure 1 shows a block form of a system in which one form of the invention may be implemented
  • Figure 2 shows the operation of the time series analysis component of Figure 1
  • Figure 3 shows a two-dimensional representation generated by the self-organising map component of Figure 1;
  • FIGS 4, 5 and 6 illustrate the application of the invention to different forms of time- variant data.
  • FIG. 1 illustrates a block diagram of the preferred system 10 in which one form of the present invention may be implemented.
  • the system could include one or more interactive client workstations 20 on which data can be displayed.
  • the system 10 includes time-variant data 30 maintained in computer memory, for example historical market data representing stock exchange trading data.
  • the data could also represent fixed interest/bonds, futures, options, foreign exchange or any other tradable instruments.
  • the time-variant data could be stored in a relational database or object-oriented database. Each stock trade is associated with a price value and a time value.
  • the database 30 is configured to enable time-variant data to be retrieved from it. Examples of retrieved data could include a series of data values representing the price of a particular stock at different time intervals.
  • the time-variant data could also include, for each trade, a trade price, a trade size, time of trade, a trade type (for example on market or over the counter), a current bid price and a current ask price.
  • the system 10 also includes a time series analysis component 40 configured to create one or more vectors from the time-variant data, as will be described below with reference to Figure 2.
  • the time series analysis component is preferably implemented as a computer program installed and operating on a computing device. Alternatively, it is envisaged that the time series analysis component could be implemented in hardware form.
  • the system 10 further includes a self-organising map (SOM) component 50 which is configured to generate and display a two-dimensional representation which includes representations of one or more of the vectors obtained from the time series analysis component 40.
  • SOM self-organising map
  • the self-organising map component 50 is preferably a software program or hardware equivalent which implements one or more artificial neural networks to map one or more of the vectors obtained from the time series analysis 40 into two dimensions.
  • a self-organising map is a neural network or neural network combination in which the parameters that are trained are a set of cell vectors that exist in the same high dimensional space as the input data vectors. Each of these represents a transformation between one cell on the self-organising map and the high dimensional space.
  • each data vector is assigned to the cell whose vector is closest to it in the high dimensional space.
  • the cell vectors are then iteratively moved around the high dimensional space so that each cell vector is moved towards the average of the vectors assigned to it and each cell vector is moved to a lesser extent towards the average of the vectors assigned to neighbouring or nearby cells. The effect of this is that the vectors mirror the data and that neighbouring cells in the two-dimensional space are near each other in the higher dimensional space.
  • a contour generator or visualisation engine 60 is configured to generate and display one or more contour lines around each vector representation as will be described below with reference to Figure 3.
  • the system 10 optionally includes a cluster analysis component 70 to determine trends or properties in the output from the self-organising map component 50.
  • the output of this cluster analysis can be exported for further analysis if required, or the properties of a cluster can be used to define a leading key performance indicator or indicators 80.
  • Market data as a real time feed 90 is transmitted to a short term data store 100 and/or a dynamic template 110.
  • Data is transmitted from the short term data store 100 and/or dynamic template 110 to a visualisation engine 120 and/or real time or near real time visualisation engine 130.
  • the visualisation engine 120 is connected to a further interactive client 140 whereas the real time visualisation engine 130 could be configured to broadcast images to real time clients 150, for example 150A, 150B and 150C.
  • Semi- static reference data for example market structured data, could also provide input to the dynamic template 110A.
  • Figure 2 illustrates a sample vector created by the time series analysis component 40 of Figure 1.
  • Raw trading data 200 is obtained from the historical market database 30 from Figure 1.
  • the time-variant data could be multivariate data, having associated with it one or more variables. These variables could include, for example, the price of shares or stocks, the volume traded and the spread in prices.
  • the time series analysis component obtains from the raw trading data values for the variables price 210, volume 220 and spread 230 at a particular time value or snapshot 240. Values of these variables at preceding time values are also obtained.
  • the resulting vector 250 represents a series of values for one or more variables, for example values 212 for price variable 210, values 222 for volume variable 220 and values 232 for spread variable 230.
  • the number of values 212 equals the number of values 222 and 232.
  • value 214 based on the value of price at a particular time value, is preferably sampled at the same time value for value 224 and value 234.
  • Values 216, 226 and 236 are preferably all sampled at the same time value but at a preceding time value to the time value associated with values 214, 224 and 234.
  • the resulting vector is a trading history vector in n-dimensional space where n can be in the hundreds or even thousands corresponding to the snapshot or current time value.
  • n can be in the hundreds or even thousands corresponding to the snapshot or current time value.
  • Every dimension can be scaled to have a range from 0 to 1 - a continuous variable z has its values z,- scaled by z; -> (z; - z m i n )/(z max - z m i n ) (where z; are the values of a dimension z across the varying data points, indexed i; z m ; n (z max ) is the minimum (maximum) of the z,).
  • a bistate or Boolean variable (male/female, true/false) is transformed to values 0,1 and a multi-state discrete variable with N+l values is transformed to (1/N) ⁇ 0,1,2,...,N ⁇ .
  • More complicated order-preserving transformations, involving the expected or known variance, or higher order modes of the variable distribution can be used to normalise the variable to a uniform, normal or other standard distribution within the unit interval.
  • FIG. 3 illustrates an example 2-dimensional representation generated by the self-organising map component.
  • Each data point represents a representation of a vector created by the time series analysis component 40.
  • the two-dimensional representation 300 for example includes data points 310, 320, 330 and 340.
  • the mapping is performed in such a way that vectors which are similar in the high dimensional space remain close together in the 2-dimensional output.
  • vectors 310 and 320 are expected to be more similar in high dimensional space due to their close proximity in the representation 30, whereas vectors 310 and 340 are expected to be more dissimilar due to the spacing between these vectors in the representation 300.
  • the visualisation engine 60 and/or the self-organising map component 50 is configured to calculate, for each vector, the price movement occurring in a given time interval after the snapshot or current time associated with each vector.
  • the absolute value of price movement is a numerical value which is associated with the individual data points representing individual vectors.
  • Vectors 310 and 320 have no associated price movement and are shown as such in the representation.
  • Vector 330 has some price movement and vector 340 has a larger price movement.
  • the visualisation engine/contour generator treats the price movements associated with each data point as a maximum value and generates and displays one or more contour lines around each data point representing a vector. Each contour line represents data values which are less than the price movement associated with the data point around which the contour line is displayed.
  • the preferred form technique for contouring is described in patent specification WO 00/77682 to Compudigm International Limited entitled "Data Visualisation System and Method".
  • contouring in this manner enables a viewer to rapidly determine patterns which precede consistent price movements. If certain clusters show strong subsequent price movements for all the trading history vector within, then that cluster represents a trading pattern that is a powerful predictive tool.
  • the viewer may then use a cluster analysis tool 70 to determine the properties of a particular cluster under consideration, resulting in a formula that can produce, for any trading history vector, a value that represents how strongly that vector would belong to the selective cluster.
  • a leading indicator For each cluster that shows strong subsequent price movements, a leading indicator is defined. For example, the Euclidean distance between a new data point and the centre of the cluster in the original high dimensional space produces a value that can be plotted. Alternatives include setting thresholds on each dimension or more sophisticated techniques such as analysis of variance (ANOVA).
  • ANOVA analysis of variance
  • the recent trading history of each stock can be turned into history vectors.
  • the system can assign each stock a number indicating the tendency of that stock to move up or down in price in the near future.
  • this leading indicator in real time, either on traditional charts or on a contoured visualisation of the entire market, traders will gain a valuable tool to assist their trading decisions.
  • the invention has several advantages. It does not attempt to replace human intuition, like other artificial neural network-based trading tools, but instead amplifies intuition with visualisations.
  • the contours show patterns more strongly than other visualisation techniques.
  • the invention combines multiple variables into a single pattern represented as the history vector.
  • the invention can also be adapted to work on many different sorts of time-variant data.
  • the invention provides an advance upon existing technology as it is a unique hybrid of time series transformation, self-organising maps, statistical clustering and contouring techniques.
  • the invention is also able to transform time-variant multivariate data into history vectors suitable for input to the self-organising map.
  • the invention is also interactive, enabling analysts to experiment with variable selection and weightings.
  • the invention could be applied to any time- variant data in addition to financial market data.
  • the time-variant data could include revenues from gaming machines in a casino context.
  • machines are grouped by similarity into clusters of a particular nature, in this case denomination, and sub-clusters for example spend patterns.
  • a contoured representation identifying a key performance indicator, for example turnover is overlaid to show performance of groupings based on the customer base of the organisation or a segment of interest within it.
  • Figure 5 illustrates the organisation of time- variant data into a series of layers including mesh block cluster boundary layers or mesh block data layers.
  • This display could also include a contour layer of market opportunity or any key performance indicator selected for each mesh block, and a thematic layer showing particular mesh blocks of interest, for example high value, low value, people that buy a particular product, or any useful profile per mesh block.
  • the projection could also include a selection layer which indicates the newly selected mesh blocks.
  • Figure 6 illustrates the invention applied to data in order to provide knowledge management.
  • the knowledge management application relies on starting with a corpus (group) of documents.
  • the documents are put through a lexical analyser system which finds a lexicon comprising all the non-trivial words (or stems or lexical roots of words) in the corpus and assigns a value to each ⁇ document, word/stem ⁇ pair.
  • the value of each pair is the importance of that word in that document, e.g., 0 if the word does not appear, Maximum if it is in the title and occurs in every paragraph. Maximum if the word is the distinguishing (from the remainder of the corpus) topic of the document.
  • the set of values for one document - one for each word in the lexicon - constitutes the vector for that document that is fed. to the self-organising map.
  • the full input is one vector for each document. Note that each dimension is of the same type as they all measure the same type of relationship.

Abstract

The invention provides a data visualisation system comprising time-variant data maintained in computer memory; a time series analysis component configured to create one or more vectors from time-variant data; and a self-organising map component configured to generate and display a two-dimensional representation including one or more vector representations. The data visualisation system also includes a contour generator configured to generate and display one or more contour lines around each vector representation. The invention also provides a related method of data visualisation.

Description

DATA VISUALISATION SYSTEM AND METHOD
FIELD OF INVENTION
The invention relates to a data visualisation system and method, particularly but not solely designed to identify patterns in time-variant data, for example stock exchange trading data. The invention is particularly suitable in identifying patterns in time- variant multivariate data.
BACKGROUND TO INVENTION
One of the biggest challenges in the financial markets is managing and interpreting the huge amounts of data and the limitation of existing tools to handle that data. Data associated with financial markets is usually time-variant. The price of stocks and shares is associated with a particular time interval, and significant price movements, both upward and downward can occur at different time intervals.
It is particularly important to identify patterns in price movements over time in order to predict short term future movements in the data. Such movements generally depend on several variables which include price, volume and spread and are particularly difficult to identify.
SUMMARY OF INVENTION
In broad terms in one form the invention comprises a data visualisation system comprising time-variant data maintained in computer memory; a time series analysis component configured to create one or more vectors from the time-variant data; a self- organising map component configured to generate and display a two-dimensional representation including one or more vector representations; and a contour generator configured to generate and display one or more contour lines around each vector representation. In broad terms in another form the invention comprises a method of data visualisation comprising the steps of maintaining time-variant data in computer memory; creating one or more vectors from the time-variant data; generating and displaying a two-dimensional representation including one or more vector representations; and generating and displaying one or more contour lines around each vector representation.
BRIEF DESCRIPTION OF THE FIGURES
Preferred forms of the data visualisation system and method will now be described with reference to the accompanying figures in which:
Figure 1 shows a block form of a system in which one form of the invention may be implemented;
Figure 2 shows the operation of the time series analysis component of Figure 1;
Figure 3 shows a two-dimensional representation generated by the self-organising map component of Figure 1; and
Figures 4, 5 and 6 illustrate the application of the invention to different forms of time- variant data.
DETAILED DESCRIPTION OF PREFERRED FORMS
Figure 1 illustrates a block diagram of the preferred system 10 in which one form of the present invention may be implemented. The system could include one or more interactive client workstations 20 on which data can be displayed. The system 10 includes time-variant data 30 maintained in computer memory, for example historical market data representing stock exchange trading data. The data could also represent fixed interest/bonds, futures, options, foreign exchange or any other tradable instruments. The time-variant data could be stored in a relational database or object-oriented database. Each stock trade is associated with a price value and a time value. The database 30 is configured to enable time-variant data to be retrieved from it. Examples of retrieved data could include a series of data values representing the price of a particular stock at different time intervals. The time-variant data could also include, for each trade, a trade price, a trade size, time of trade, a trade type (for example on market or over the counter), a current bid price and a current ask price.
The system 10 also includes a time series analysis component 40 configured to create one or more vectors from the time-variant data, as will be described below with reference to Figure 2. The time series analysis component is preferably implemented as a computer program installed and operating on a computing device. Alternatively, it is envisaged that the time series analysis component could be implemented in hardware form.
The system 10 further includes a self-organising map (SOM) component 50 which is configured to generate and display a two-dimensional representation which includes representations of one or more of the vectors obtained from the time series analysis component 40. The self-organising map component 50 is preferably a software program or hardware equivalent which implements one or more artificial neural networks to map one or more of the vectors obtained from the time series analysis 40 into two dimensions.
A self-organising map is a neural network or neural network combination in which the parameters that are trained are a set of cell vectors that exist in the same high dimensional space as the input data vectors. Each of these represents a transformation between one cell on the self-organising map and the high dimensional space. To train the self- organising map, each data vector is assigned to the cell whose vector is closest to it in the high dimensional space. The cell vectors are then iteratively moved around the high dimensional space so that each cell vector is moved towards the average of the vectors assigned to it and each cell vector is moved to a lesser extent towards the average of the vectors assigned to neighbouring or nearby cells. The effect of this is that the vectors mirror the data and that neighbouring cells in the two-dimensional space are near each other in the higher dimensional space.
A contour generator or visualisation engine 60 is configured to generate and display one or more contour lines around each vector representation as will be described below with reference to Figure 3.
The system 10 optionally includes a cluster analysis component 70 to determine trends or properties in the output from the self-organising map component 50.
The output of this cluster analysis can be exported for further analysis if required, or the properties of a cluster can be used to define a leading key performance indicator or indicators 80.
Market data as a real time feed 90 is transmitted to a short term data store 100 and/or a dynamic template 110. Data is transmitted from the short term data store 100 and/or dynamic template 110 to a visualisation engine 120 and/or real time or near real time visualisation engine 130. The visualisation engine 120 is connected to a further interactive client 140 whereas the real time visualisation engine 130 could be configured to broadcast images to real time clients 150, for example 150A, 150B and 150C. Semi- static reference data, for example market structured data, could also provide input to the dynamic template 110A.
Figure 2 illustrates a sample vector created by the time series analysis component 40 of Figure 1. Raw trading data 200 is obtained from the historical market database 30 from Figure 1. The time-variant data could be multivariate data, having associated with it one or more variables. These variables could include, for example, the price of shares or stocks, the volume traded and the spread in prices. The time series analysis component obtains from the raw trading data values for the variables price 210, volume 220 and spread 230 at a particular time value or snapshot 240. Values of these variables at preceding time values are also obtained. The resulting vector 250 represents a series of values for one or more variables, for example values 212 for price variable 210, values 222 for volume variable 220 and values 232 for spread variable 230.
Preferably, the number of values 212 equals the number of values 222 and 232. Furthermore, value 214, based on the value of price at a particular time value, is preferably sampled at the same time value for value 224 and value 234. Values 216, 226 and 236 are preferably all sampled at the same time value but at a preceding time value to the time value associated with values 214, 224 and 234.
Only three variables are shown in Figure 2 for simplicity, and it will be appreciated that the number of actual variables could be many times this number. Furthermore, only nine different time values are shown for simplicity and it will be appreciated that the number of time values in practice could be many times this number.
The resulting vector is a trading history vector in n-dimensional space where n can be in the hundreds or even thousands corresponding to the snapshot or current time value. By sampling thousands of such snapshots of different stocks at different times, the time series analysis component is able to create a large space of high-dimensional vectors. Within this space, there are likely to be significant structures, representing recurrent patterns in the trading data, but the vast dimensionality of the space makes it impossible for humans to discover these structures unaided.
Where necessary, the data in one or more of the resulting vectors is transformed, scaled and statistically normalised to ensure that comparisons between the often disparate variables are meaningful. Every dimension can be scaled to have a range from 0 to 1 - a continuous variable z has its values z,- scaled by z; -> (z; - zmin)/(zmax - zmin) (where z; are the values of a dimension z across the varying data points, indexed i; zm;n (zmax) is the minimum (maximum) of the z,). A bistate or Boolean variable (male/female, true/false) is transformed to values 0,1 and a multi-state discrete variable with N+l values is transformed to (1/N) {0,1,2,...,N}. More complicated order-preserving transformations, involving the expected or known variance, or higher order modes of the variable distribution can be used to normalise the variable to a uniform, normal or other standard distribution within the unit interval.
The vector or vectors resulting from the time series analysis component 40 is then passed to the self-organising map component 50. Figure 3 illustrates an example 2-dimensional representation generated by the self-organising map component. Each data point represents a representation of a vector created by the time series analysis component 40. The two-dimensional representation 300 for example includes data points 310, 320, 330 and 340. Preferably the mapping is performed in such a way that vectors which are similar in the high dimensional space remain close together in the 2-dimensional output. On this basis, vectors 310 and 320 are expected to be more similar in high dimensional space due to their close proximity in the representation 30, whereas vectors 310 and 340 are expected to be more dissimilar due to the spacing between these vectors in the representation 300.
The visualisation engine 60 and/or the self-organising map component 50 is configured to calculate, for each vector, the price movement occurring in a given time interval after the snapshot or current time associated with each vector. The absolute value of price movement is a numerical value which is associated with the individual data points representing individual vectors. Vectors 310 and 320 have no associated price movement and are shown as such in the representation. Vector 330 has some price movement and vector 340 has a larger price movement.
The visualisation engine/contour generator treats the price movements associated with each data point as a maximum value and generates and displays one or more contour lines around each data point representing a vector. Each contour line represents data values which are less than the price movement associated with the data point around which the contour line is displayed. The preferred form technique for contouring is described in patent specification WO 00/77682 to Compudigm International Limited entitled "Data Visualisation System and Method".
The use of contouring in this manner enables a viewer to rapidly determine patterns which precede consistent price movements. If certain clusters show strong subsequent price movements for all the trading history vector within, then that cluster represents a trading pattern that is a powerful predictive tool.
The viewer may then use a cluster analysis tool 70 to determine the properties of a particular cluster under consideration, resulting in a formula that can produce, for any trading history vector, a value that represents how strongly that vector would belong to the selective cluster.
For each cluster that shows strong subsequent price movements, a leading indicator is defined. For example, the Euclidean distance between a new data point and the centre of the cluster in the original high dimensional space produces a value that can be plotted. Alternatives include setting thresholds on each dimension or more sophisticated techniques such as analysis of variance (ANOVA).
Referring to Figure 1, as live trading data 90 arrives from a data feed, the recent trading history of each stock can be turned into history vectors. By applying the leading indicator definition accordingly, the system can assign each stock a number indicating the tendency of that stock to move up or down in price in the near future. By plotting this leading indicator in real time, either on traditional charts or on a contoured visualisation of the entire market, traders will gain a valuable tool to assist their trading decisions.
The invention has several advantages. It does not attempt to replace human intuition, like other artificial neural network-based trading tools, but instead amplifies intuition with visualisations. The contours show patterns more strongly than other visualisation techniques. The invention combines multiple variables into a single pattern represented as the history vector. The invention can also be adapted to work on many different sorts of time-variant data.
The invention provides an advance upon existing technology as it is a unique hybrid of time series transformation, self-organising maps, statistical clustering and contouring techniques. The invention is also able to transform time-variant multivariate data into history vectors suitable for input to the self-organising map. The invention is also interactive, enabling analysts to experiment with variable selection and weightings.
It is envisaged that the invention could be applied to any time- variant data in addition to financial market data. In one form the time-variant data could include revenues from gaming machines in a casino context. Referring to Figure 4, machines are grouped by similarity into clusters of a particular nature, in this case denomination, and sub-clusters for example spend patterns. Onto these groupings, a contoured representation identifying a key performance indicator, for example turnover, is overlaid to show performance of groupings based on the customer base of the organisation or a segment of interest within it.
Figure 5 illustrates the organisation of time- variant data into a series of layers including mesh block cluster boundary layers or mesh block data layers. This display could also include a contour layer of market opportunity or any key performance indicator selected for each mesh block, and a thematic layer showing particular mesh blocks of interest, for example high value, low value, people that buy a particular product, or any useful profile per mesh block. The projection could also include a selection layer which indicates the newly selected mesh blocks.
Figure 6 illustrates the invention applied to data in order to provide knowledge management. The knowledge management application relies on starting with a corpus (group) of documents. The documents are put through a lexical analyser system which finds a lexicon comprising all the non-trivial words (or stems or lexical roots of words) in the corpus and assigns a value to each {document, word/stem} pair. The value of each pair is the importance of that word in that document, e.g., 0 if the word does not appear, Maximum if it is in the title and occurs in every paragraph. Maximum if the word is the distinguishing (from the remainder of the corpus) topic of the document.
Typically words that have a high rating in nearly all documents (company name) are ignored as their discriminating power is limited. The set of values for one document - one for each word in the lexicon - constitutes the vector for that document that is fed. to the self-organising map. The full input is one vector for each document. Note that each dimension is of the same type as they all measure the same type of relationship.
The foregoing describes the invention including preferred forms thereof. Alterations and modifications as will be obvious to those skilled in the art are intended to be incorporated within the scope hereof, as defined by the accompanying claims.

Claims

CLAIMS:
1. A data visualisation system comprising: time-variant data maintained in computer memory; a time series analysis component configured to create one or more vectors from the time-variant data; a self-organising map component configured to generate and display a two- dimensional representation including one or more vector representations; and a contour generator configured to generate and display one or more contour lines around each vector representation.
2. A data visualisation system as claimed in claim 1 wherein the time series analysis component is configured to create at least one vector based on the value of a variable at a time value, and a series of values of the variable at preceding time values.
3. A data visualisation system as claimed in claim 2 wherein the time series analysis component is configured to create at least one vector based on the values of a plurality of variables at a time value, and a series of values of the variable at preceding time values.
4. A method of data visualisation comprising the steps of: maintaining time-variant data in computer memory; creating one or more vectors from the time- variant data; generating and displaying a two-dimensional representation including one or more vector representations; and generating and displaying one or more contour lines around each vector representation.
5. A method of data visualisation as claimed in claim 4 wherein the step of creating one or more vectors is based on the value of a variable at a time value, and a series of values of the variable at preceding time values .
6. A method of data visualisation as claimed in claim 5 further comprising the step of creating at least one vector based on the values of a plurality of variables at a time value, and a series of values of the variable at preceding time values.
PCT/NZ2002/000138 2001-07-30 2002-07-30 Data visualisation system and method WO2003012713A1 (en)

Applications Claiming Priority (6)

Application Number Priority Date Filing Date Title
NZ513265 2001-07-30
NZ51326501 2001-07-30
NZ51638201 2001-12-24
NZ516382 2001-12-24
NZ519832 2002-06-27
NZ51983202 2002-06-27

Publications (1)

Publication Number Publication Date
WO2003012713A1 true WO2003012713A1 (en) 2003-02-13

Family

ID=27353939

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/NZ2002/000138 WO2003012713A1 (en) 2001-07-30 2002-07-30 Data visualisation system and method

Country Status (1)

Country Link
WO (1) WO2003012713A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2395807A (en) * 2002-11-27 2004-06-02 Sony Uk Ltd Information retrieval
GB2395805A (en) * 2002-11-27 2004-06-02 Sony Uk Ltd Information retrieval
CN1808414B (en) * 2004-12-06 2010-05-26 索尼株式会社 Method and apparatus for recognizing observation value of time sequence data
US8235811B2 (en) 2007-03-23 2012-08-07 Wms Gaming, Inc. Using player information in wagering game environments

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6035057A (en) * 1997-03-10 2000-03-07 Hoffman; Efrem H. Hierarchical data matrix pattern recognition and identification system
WO2001004808A1 (en) * 1999-07-13 2001-01-18 Compudigm International Limited Interaction prediction system and method
US6301579B1 (en) * 1998-10-20 2001-10-09 Silicon Graphics, Inc. Method, system, and computer program product for visualizing a data structure
US6307500B1 (en) * 1999-08-13 2001-10-23 University Corporation For Atmospheric Research Method of moment estimation and feature extraction for devices which measure spectra as a function of range or time

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6035057A (en) * 1997-03-10 2000-03-07 Hoffman; Efrem H. Hierarchical data matrix pattern recognition and identification system
US6301579B1 (en) * 1998-10-20 2001-10-09 Silicon Graphics, Inc. Method, system, and computer program product for visualizing a data structure
WO2001004808A1 (en) * 1999-07-13 2001-01-18 Compudigm International Limited Interaction prediction system and method
US6307500B1 (en) * 1999-08-13 2001-10-23 University Corporation For Atmospheric Research Method of moment estimation and feature extraction for devices which measure spectra as a function of range or time

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2395807A (en) * 2002-11-27 2004-06-02 Sony Uk Ltd Information retrieval
GB2395805A (en) * 2002-11-27 2004-06-02 Sony Uk Ltd Information retrieval
US7668853B2 (en) 2002-11-27 2010-02-23 Sony United Kingdom Limited Information storage and retrieval
CN1808414B (en) * 2004-12-06 2010-05-26 索尼株式会社 Method and apparatus for recognizing observation value of time sequence data
US8235811B2 (en) 2007-03-23 2012-08-07 Wms Gaming, Inc. Using player information in wagering game environments
US9619969B2 (en) 2007-03-23 2017-04-11 Bally Gaming, Inc. Using player information in wagering game environments

Similar Documents

Publication Publication Date Title
Kovacova et al. Systematic review of variables applied in bankruptcy prediction models of Visegrad group countries
Matsunaga et al. Exploring graph neural networks for stock market predictions with rolling window analysis
Ziegler et al. Visual market sector analysis for financial time series data
Bagnall et al. Clustering time series with clipped data
Owsley et al. Self-organizing feature maps and hidden Markov models for machine-tool monitoring
Sarlin et al. Visual predictions of currency crises using self-organizing maps
US5897627A (en) Method of determining statistically meaningful rules
CA2480954A1 (en) Viewing multi-dimensional data through hierarchical visualization
Rathnayaka et al. Geometric Brownian motion with Ito's lemma approach to evaluate market fluctuations: A case study on Colombo Stock Exchange
Durica et al. Business failure prediction using cart-based model: A case of Slovak companies.
Mendes et al. Reconstructing an economic space from a market metric
Kamani et al. Data Normalization in Data Mining Using Graphical User Interface a Pre-Processing Stage
WO2003012713A1 (en) Data visualisation system and method
Dybowski et al. Visualization of binary string convergence by sammon mapping
Condrobimo et al. Data mining technique with cluster anaysis use K-means algorithm for LQ45 index on Indonesia stock exchange
Andriyashin Financial Applications ofClassification and Regression Trees
Supriya et al. House Price Prediction System using Machine Learning Algorithms and Visualization
Abdelatif et al. Optimization of the organized Kohonen map by a new model of preprocessing phase and application in clustering
Brandl et al. An automated econometric decision support system: forecasts for foreign exchange trades
Shrivastava et al. Corporate distress prediction using random forest and tree net for india
Gaing et al. Optimal grey topological predicting approach to short-term load forecasting in power system
Chen On a machine learning based analysis of online transaction
Pitera An assessment of the reliability of early warning models on the example of small and medium-sized enterprises in the industry and services sector
KR20200048641A (en) Parts management device and operating method of the same
CN111597505B (en) Correlation analysis method and correlation device for electricity utilization users in power network

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BY BZ CA CH CN CO CR CU CZ DE DM EC EE ES FI GB GD GE GH GM HU ID IL IN IS JP KE KG KP KR KZ LK LR LS LT LU LV MA MD MG MK MW MX MZ NO NZ OM PH PL PT RO SD SE SG SI SK SL TJ TM TN TR TT UA UG US UZ VN YU ZA ZM

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM EC EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NO NZ OM PH PL PT RO RU SD SE SG SI SK SL TJ TM TN TR TT TZ UA UG US UZ VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): GH GM KE LS MW MZ SD SL SZ UG ZM ZW AM AZ BY KG KZ RU TJ TM AT BE BG CH CY CZ DK EE ES FI FR GB GR IE IT LU MC PT SE SK TR BF BJ CF CG CI GA GN GQ GW ML MR NE SN TD TG

Kind code of ref document: A1

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR IE IT LU MC NL PT SE SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
REG Reference to national code

Ref country code: DE

Ref legal event code: 8642

122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase

Ref country code: JP

WWW Wipo information: withdrawn in national office

Country of ref document: JP