Patent Grant
7640198
Price modeling using a variety of analytical methods has been described in many publications including, for example, economic treatises, textbooks, and patents. Numerical information culled from vast data sets containing numerous transaction based sales operations may be analyzed and displayed in any of a number of different ways such as through word processing, spreadsheets, and graphical software programs. Modeling this numerical information visually in a price modeling context presents various challenges to designers and analysts.
In order to effectively communicate via visual display methods information gleaned from data sets, designers face many challenges. In one instance the sheer number of data entries in a transaction based data warehouse can typically exceed many millions of transactions. Displaying millions of individual entries on a single display will not generally provide an analyst with useful information. That is, simply graphing a large number of entries alone will not generally provide useful insight into the characteristic and nature of the data set. In this situation, predictive analysis of the data becomes difficult, if not impossible, as a practical matter. Thus, designers must, in some fashion, aggregate, or allow the user to flexibly aggregate the data in order to display any meaningful characterization of a data set.
Typically, generally known statistical analysis methods may be employed to aggregate data. Averages, medians, modes, and other statistical methods well known in the art may be utilized to aggregate data so that trends and analysis may be affected. However, in some cases, unique methods of data aggregation may be desirable. For example, an average value of price quotes for a particular region may give a sales person a basis upon which to make a real time price quote to a customer. The average value, in this example, represents a statistical locus around which all the price quotes for that region tend to fall. Averaging has an added benefit of reducing a chosen set of data to a single number thus deriving performance gains when analyzing and manipulating the data set. However, as can be appreciated by one skilled in the art, average values cannot generally account for relative comparisons between groups of related items. To make a relative comparison between groups of related items, an index may be calculated.
Indexes, which are generally known in the art, have been employed in a variety of manners. Stock markets, for example, often use indexes as a gauge of general market condition. Index calculations are typically performed in batch processes. As such, indexes are indicators of past performance only and generally cannot be used to compare real-time data changes. Indexing also tends to aggregate data in ways that make it difficult if not impossible to explore lower level data orders such as individual transactions for example. Further, while indexes generally provide a single value as an indicator, they are not generally visualized in other ways. Thus, an innovative index calculation that may be generated in response to real-time data changes, that allows extraction and manipulation of the underlying data, and that may be visualized in new ways is desirable to achieve a more robust and rich analysis.
Another complication confronting designers in modeling and displaying large transaction based data sets is that in typical legacy systems, aggregating data often results in the loss of individual transaction information due in part to methodologies selected to enhance performance. As noted above, an average value represents a statistical locus around which all values in the data tend to fall. Also noted above, averaging represents a way of reducing a data set. In many cases, however, a finer level of granularity with respect to the data set may be desirable since, in at least some instances, outliers and dispersions of data tend to become aggregated or averaged away resulting in a distortion of original data set. Thus, for example, a data set may be reduced by averaging based on a given criteria (e.g., average prices for a given item by region). The data set, in this example, may be reduced to a single entry for each item in a given region. As an example, the average price of a widget in the western region may be represented as a single entry. Unfortunately, however, once a data set is reduced, the ability to examine a single transaction or even a group of transactions within the data set is typically severely limited or impossible. Thus, methods to recapture and display single or group transactions of a reduced data set may be desirable.
In view of the foregoing, systems and methods for generating and displaying indexed price modeling data.
The present invention presents systems and methods for suitable generating and displaying indexed price modeling data. The present invention allows a user to perform index calculations in real-time and to display resulting indexes. Indexes may be summarized according to user preferences and mined to extract underlying data.
One embodiment of the present invention provides a system suitable for displaying price modeling data having an indexing module that calculates indexes based on price modeling data and a display module configured display a calculated index. Another embodiment provides a data aggregation module that aggregates indexes calculated by the indexing module that may be also be displayed.
Another embodiment of the present invention presents a method for suitably displaying price modeling data. In the method, price modeling data is used to calculate an index. Price modeling data may also be used to create an aggregate index based on user selections in accordance with the present invention. The calculated index and the aggregate index may be displayed in any combination.
In still other embodiments of the present invention provides a computer program product in a computer readable media suitable for displaying price modeling data. The computer program product includes an index calculation module that calculates indexes based on price modeling data and a display module configured to display calculated indexes.
Embodiments of the invention may best be understood by reference to the following description taken in conjunction with the accompanying drawings in which:
Once commands are executed by a server or servers, results may be sent to a client in XML at step 116. XML does not return transactional data to be processed and displayed by a client. Rather, as noted above, the command computations are executed at a server level which then returns XML to a client to display computational results. In this manner, results may be displayed by any method capable of displaying XML data. The described invention, however, is not limited to sending and receiving XML data and may thus be implemented using any suitable network communication protocol. Results sent in step 116 may be received by a client at step 120 and displayed by a client at step 124. In the example embodiment provided, any XML capable browser may display results received at step 120.
Once results are displayed, a client determines whether more commands are to be processed at step 128. If more commands need to be processed, the method returns to step 100 and processes commands as described above. When all commands have been completed, a user may sign off and the method ends.
calculated value=entry value−mean value Equation 1: Example Calculated Value
The resulting calculated value may then be stored in a new data structure along with imported data. Any number of user defined calculation parameters may be incorporated into a client configuration file. In one embodiment of the present invention, an index is a calculation parameter. Index generation will be discussed in further detail for
After a client configuration file is read, the method determines whether a transaction table all ready exists at step 204. Transaction tables, at a general level, represent the data extracted from the data source in table format along with calculated values. More particularly, a transaction table is a two dimensional table with rows and columns. Each row, in some embodiments, represent a transaction while each column represents some characteristic, description, or calculated value corresponding to a row. It may be appreciated that the designation of rows and columns is for convenience only since a transaction table may also be organized as columns representing transactions and rows representing characteristics, descriptions, or calculated values corresponding to a column.
Thus, for example, a transaction table may be populated as follows:
In the above example, a transaction table is populated with four transactions and four columns of data corresponding to each transaction. Some of the entries may be defined as measurements which are typically numeric and others may be defined as dimensions which are typically descriptive elements of the transaction. It is noted that this table reflects actual values for the given fields; however, in some embodiments the values may be converted into expression objects which have the advantage of being syntactically equivalent so that operations on both measurements and definitions may be similarly processed. It is also noted that the above table contains only four transaction entries and five descriptive fields that are either measurements or dimensions. The present invention contemplates many more transactions and many more columns. In a preferred embodiment, the table may contain up to approximately three million records. In other preferred embodiments, the table may contain up to approximately one million records. Memory limitations described in these embodiments are due in part to the limitations associated with 32-bit architecture currently supported by most IT organizations. However, this limitation may be overcome by porting the present invention to a higher capacity 64-bit architecture and above.
If a transaction table is found at step 204, the method purges a found transaction table from memory at step 208. In this manner, old transaction tables are removed before new tables are formed thus avoiding data corruption. Once an old transaction table is purged, the method continues to step 212. If a transaction table is not found at step 204, the method continues to step 212 where data is requested from a data source in accordance with a configuration file read at step 200. A data source may be any of a number of data sources well-known in the art including, for example, Oracle databases or SAP databases. As noted above, configuration files map data source data to conform to a desired transaction table format. Once data has been extracted from a data source, extracted data is transformed at step 220 using a configuration file read at step 200. As noted above, a configuration file may contain, along with other parameters, calculation parameters to be applied to extracted data sets. For example, in Table 1 above, a calculated parameter illustrated is defined by the equation: calculated value=price−average price. Thus, for each transaction a value may be transformed by a calculation applied a transaction. In a preferred embodiment, at least one index is calculated. Indexes may be, without limitation, numeric values, or some other relative absolute value. Index calculations are discussed in further detail below for
Once extracted data is transformed, a resulting data set is loaded into memory (RAM) as a transaction table containing all desired transactions from a data source along with all transformed data generated at step 224. As can be appreciated by one skilled in the art, 32-bit platforms can only address up to 4.0 gigabytes of physical memory (RAM) (i.e., 232=4000 million). In WINDOWS™ operating system, processes are limited to 2.0 gigabytes of RAM. In some embodiments using Java Virtual Machine (JVM), the process space is further limited to 1.5 gigabytes of RAM. Thus, in a preferred embodiment, the amount of usable memory (RAM) for a transaction table is approximately 1.5 gigabytes. More preferably, the amount of usable memory (RAM) for a transaction table is approximately 1.0 gigabytes. As noted above, in systems utilizing 64-bit platforms, no such memory limitations are contemplated. After a transaction table is loaded into memory (RAM) at step 224, the system then waits for user input at as step 228 of the type described, for example, in
As noted above for
An index numerator is then read or calculated at a step 312. A numerator may be a data entry like, for example, a list price or it may be a calculated value like, for example, list price less invoice price. In either case, a numerator may be stored in a transaction table. Once an index denominator and an index numerator are calculated at steps 308 and 312 respectively, an index may be calculated at step 316 and loaded into an index column at step 320. Using the above mentioned denominator and numerator, an example of an index formulation is illustrated according to the following equation:
The method then determines whether another desired transaction exists for which an index may be calculated at step 324. If another desired transaction exists, the method returns to step 312 and cycles until all desired transactions are processed. When all transactions are processed, the method then determines whether all indexes have been processed at step 328. If more index calculations need processing, the method returns to step 304 and cycles until all index calculations have been processed. When all indexes have been processed, the method either returns to step 224 if all transformations are complete, or continues transforming data. Populating a transaction table to be loaded into memory (RAM) has been described above.
It can be appreciated that indexes, as disclosed may be calculated in real-time using a current data set. Furthermore, transforming data in accordance with the present invention does not result in the loss of information because a transaction table, upon which an index is calculated, is preserved. Thus, a user may freely explore data that underlies an index calculation, thus yielding a richer research tool. In addition, indexes, as contemplated by the present invention, may be rolled up. Generally speaking, roll up allows a user to summarize by field a set of data. In this example, because an underlying data set for a given index calculation is available; a user may select data descriptors that may further delineate an index calculation. Roll up will be discussed in further detail below for
After a user has selected a particular chart at step 404, a user may select appropriate axis parameters at a step 408. Axis parameters represent a desired data set to be plotted. For example, price indicators (y-axis parameter) may be plotted against temporal indicators (x-axis parameter) to determine the change in pricing over time. In like manner, temporal indicators (y-axis parameter) may be plotted against price indicators (x-axis parameter). Thus, selection of axis parameters may be highly flexible according to user preference. In some embodiments, indexes may be selected as axis parameters. In other embodiments, axis parameters may be selected from drop down menus that contain many possible parameters in accordance with a corresponding configuration file. Axis parameters may be further selected in any manner known in the art without limitation.
At another step 412, a user may select any of a number of different filters. Filters will be discussed in further detail below for
At another step 420, a user may select a roll up criteria. Roll up will be discussed in further detail below for
It can be appreciated that the operations described under steps 408 through 420 may be selected in any order in accordance with a user's preferences. Furthermore, selections described under steps 412 and 416 may be multiply selected and preserved in any order in accordance with a user's preferences and may be displayed in a selection list. Still further, selections may be individually or multiply added to or removed from an existing selection list. In some alternate embodiments, filter data sets may be retained in and subsequently recovered from cache. In other alternate embodiments, zoom data sets may be retained in and subsequently recovered from cache. Recovering data sets from cache may realize performance gains and memory efficiencies. Finally, in an embodiment of the present invention, a user may select any type of chart under step 404 while preserving selections made previous to a chart selection. That is, in some embodiments, axis parameters, filters, zoom, and roll up may persist across chart selections.
After filter criterion is received, an expression object based on a received filter criterion is created at a step 504. Creation of an expression object allows for efficient syntactical processes to be achieved that may result in performance advantages. Once an expression object is created for a filter at step 504, a determination is made as to whether a roll up table is required. As noted above, a roll up allows a user to aggregate data according to a selection criterion. Aggregated data may also be summarized in accordance with a desired axis parameter. Roll up requirements must be considered before a filter is applied to a data set because a rolled up data set may respond differently to a selected filter than an original data set. Roll up will be discussed in further detail below for
At a step 520, a row in a table (e.g., transaction table, or roll up table) is evaluated according to a filter expression object. A filter column corresponding to a selected filter may then be created in a table to hold a Boolean result of the evaluation of step 520 at step 524. The method then determines whether a row under examination matches a selected filter (i.e., Boolean=true) at a step 528. If a row under examination matches a selected filter, then that row is added to a row set representing a set of data matching a selected filter at a step 532. The method then determines whether more rows need evaluation at step 536. If more rows need evaluating, the method returns to step 520 and cycles until all rows in a table are evaluated. If, at step 528, a row under examination does not match a selected filter (i.e., Boolean=false), then the row under examination is not added to a row set and the method continues at step 536 to determine whether more rows need evaluation. As noted above, if more rows need evaluating, the method returns to step 520 and cycles until all rows in a table are evaluated.
When all rows have been evaluated, the method continues at a step 540 to determine whether additional filters have been selected. If additional filters have been selected, the method returns to step 500 and cycles and continues until all filters have been evaluated. The method then ends. Note that each filter selection requires roll up evaluation to assure that the roll up is properly applied in a sequence of selected operations.
At a step 600, zoom criteria selected by a user are received. As noted above, zoom criteria may be selected (by mouse drag) or manually input by a user. Typically, zooms may be organized by dimension and by measure. Dimension is an attribute of a transaction that can have one of a known list of values. For example, every transaction has customer and there is a known set of customers. Thus customer is an example of a dimension. A dimension may have a flat list of values or it may have a hierarchical list. Measure is an attribute that has a numeric value. In some embodiments, a value may be an amount of money. Thus, selection by measure is a selection based on a numeric value.
After zoom criteria are received, an expression object based on received zoom criteria is created at a step 604. As noted above, creation of an expression object allows for efficient syntactical processes to be achieved that may result in performance advantages. Once an expression object is created for a zoom at step 604, a determination is made as to whether a roll up table is required. As noted above, a roll up allows a user to aggregate data according to a selection criterion. Aggregated data may also be summarized in accordance with a desired axis parameter. Roll up requirements must be considered before a zoom is applied to a data set because a resulting roll up data set may respond differently to a selected zoom than an original data set. Roll up will be discussed in further detail below for
At a step 620, a row in the table (e.g., transaction table, or roll up table) is evaluated according to a zoom expression object. A zoom column corresponding to a selected zoom is then created in a table to hold a Boolean result of the evaluation of step 620 at step 624. The method then determines whether a row under examination matches a selected zoom (i.e., Boolean=true) at a step 628. If a row under examination matches a selected zoom, that row under examination is added to a row set representing a set of data matching a selected zoom at a step 632. The method then determines whether more rows need evaluation at step 636. If more rows need evaluating, the method returns to step 620 and cycles until all rows in a table are evaluated. If, at step 628, a row under examination does not match a selected zoom (i.e., Boolean=false), then that row under examination is not added to a row set and the method continues at step 636 to determine whether more rows need evaluation. As noted above, if more rows need evaluating, the method returns to step 620 and cycles until all rows in a table are evaluated.
When all rows have been evaluated, the method continues at a step 640 to determine whether additional zooms have been selected. If additional zooms have been selected, the method returns to step 600 and cycles and continues until all zooms have been evaluated. The method then ends. Note that each zoom selection requires roll up evaluation to assure that a roll up is properly applied in a sequence of selected operations. Note that the use of an expression object allows for substantially identical syntactical processing and may result in more efficient code.
At a step 700, a roll up selection criterion is read. Roll up selection criterion may be selected by menu or by input as desired by a user. The method continues at a step 704 to determine whether a roll up table matching a roll up selection criterion is available in cache. By using cached tables, the method may achieve performance advantages over prior art methodologies. If a roll up table exists in cache, a cached roll up table is read into memory at a step 720 whereupon the method ends. If a roll up table is not available, the method reads a transaction table at a step 708 that was created at step 224,
While this invention has been described in terms of several preferred embodiments, there are alterations, permutations, modifications and various substitute equivalents, which fall within the scope of this invention. For example, the portfolios illustrated in
| Number | Name | Date | Kind |
|---|---|---|---|
| 3806711 | Cousins, Jr. | Apr 1974 | A |
| 5053957 | Suzuki | Oct 1991 | A |
| 5224034 | Katz et al. | Jun 1993 | A |
| 5461708 | Kahn | Oct 1995 | A |
| 5497489 | Menne | Mar 1996 | A |
| 5537590 | Amado | Jul 1996 | A |
| 5590269 | Kruse et al. | Dec 1996 | A |
| 5670984 | Robertson et al. | Sep 1997 | A |
| 5689287 | Mackinlay et al. | Nov 1997 | A |
| 5710887 | Chelliah et al. | Jan 1998 | A |
| 5740448 | Gentry et al. | Apr 1998 | A |
| 5758327 | Gardner et al. | May 1998 | A |
| 5808894 | Wiens et al. | Sep 1998 | A |
| 5870717 | Wiecha | Feb 1999 | A |
| 5873069 | Reuhl et al. | Feb 1999 | A |
| 5878400 | Carter, III | Mar 1999 | A |
| 5946666 | Nevo et al. | Aug 1999 | A |
| 6009407 | Garg | Dec 1999 | A |
| 6075530 | Lucas et al. | Jun 2000 | A |
| 6078901 | Ching | Jun 2000 | A |
| 6151031 | Atkins et al. | Nov 2000 | A |
| 6211880 | Impink, Jr. | Apr 2001 | B1 |
| 6320586 | Plattner et al. | Nov 2001 | B1 |
| 6434533 | Fitzgerald | Aug 2002 | B1 |
| 6553350 | Carter | Apr 2003 | B2 |
| 6665577 | Onyshkevych et al. | Dec 2003 | B2 |
| 6678695 | Bonneau et al. | Jan 2004 | B1 |
| 6785664 | Jameson | Aug 2004 | B2 |
| 6801201 | Escher | Oct 2004 | B2 |
| 6812926 | Rugge | Nov 2004 | B1 |
| 6851604 | Girotto et al. | Feb 2005 | B2 |
| 6856967 | Woolston et al. | Feb 2005 | B1 |
| 6907403 | Klein et al. | Jun 2005 | B1 |
| 6988076 | Ouimet | Jan 2006 | B2 |
| 7015912 | Marais | Mar 2006 | B2 |
| 7046248 | Perttunen | May 2006 | B1 |
| 7076463 | Boies et al. | Jul 2006 | B1 |
| 7080026 | Singh et al. | Jul 2006 | B2 |
| 7092929 | Dvorak et al. | Aug 2006 | B1 |
| 7133848 | Phillips et al. | Nov 2006 | B2 |
| 7149716 | Gatto | Dec 2006 | B2 |
| 7155510 | Kaplan | Dec 2006 | B1 |
| 7218325 | Buck | May 2007 | B1 |
| 7233928 | Huerta et al. | Jun 2007 | B2 |
| 7254584 | Addison, Jr. | Aug 2007 | B1 |
| 7308421 | Raghupathy et al. | Dec 2007 | B2 |
| 7315835 | Takayasu et al. | Jan 2008 | B1 |
| 7343355 | Ivanov et al. | Mar 2008 | B2 |
| 20010003814 | Hirayama et al. | Jun 2001 | A1 |
| 20020007323 | Tamatsu | Jan 2002 | A1 |
| 20020032610 | Gold et al. | Mar 2002 | A1 |
| 20020042782 | Albazz et al. | Apr 2002 | A1 |
| 20020052817 | Dines et al. | May 2002 | A1 |
| 20020059229 | Natsumeda et al. | May 2002 | A1 |
| 20020072993 | Sandus et al. | Jun 2002 | A1 |
| 20020099596 | Geraghty | Jul 2002 | A1 |
| 20020107819 | Ouimet | Aug 2002 | A1 |
| 20020116348 | Phillips et al. | Aug 2002 | A1 |
| 20020128953 | Quallen et al. | Sep 2002 | A1 |
| 20020152133 | King et al. | Oct 2002 | A1 |
| 20020152150 | Cooper et al. | Oct 2002 | A1 |
| 20020156695 | Edwards | Oct 2002 | A1 |
| 20020165726 | Grundfest | Nov 2002 | A1 |
| 20020165760 | Delurgio et al. | Nov 2002 | A1 |
| 20020178077 | Katz et al. | Nov 2002 | A1 |
| 20020188576 | Peterson et al. | Dec 2002 | A1 |
| 20020194051 | Hall et al. | Dec 2002 | A1 |
| 20030028451 | Ananian | Feb 2003 | A1 |
| 20030033240 | Balson et al. | Feb 2003 | A1 |
| 20030095256 | Cargill et al. | May 2003 | A1 |
| 20030110066 | Walser et al. | Jun 2003 | A1 |
| 20030126053 | Boswell et al. | Jul 2003 | A1 |
| 20030130883 | Schroeder et al. | Jul 2003 | A1 |
| 20030167209 | Hsieh | Sep 2003 | A1 |
| 20030191723 | Foretich et al. | Oct 2003 | A1 |
| 20030195810 | Raghupathy et al. | Oct 2003 | A1 |
| 20030200185 | Huerta et al. | Oct 2003 | A1 |
| 20030225593 | Ternoey et al. | Dec 2003 | A1 |
| 20030229552 | Lebaric et al. | Dec 2003 | A1 |
| 20040024715 | Ouimet | Feb 2004 | A1 |
| 20040049470 | Ouimet | Mar 2004 | A1 |
| 20040078288 | Forbis et al. | Apr 2004 | A1 |
| 20040117376 | Lavin et al. | Jun 2004 | A1 |
| 20040128225 | Thompson et al. | Jul 2004 | A1 |
| 20040133526 | Shmueli et al. | Jul 2004 | A1 |
| 20040193442 | Kimata et al. | Sep 2004 | A1 |
| 20040267674 | Feng et al. | Dec 2004 | A1 |
| 20050004819 | Etzioni et al. | Jan 2005 | A1 |
| 20050096963 | Myr et al. | May 2005 | A1 |
| 20050197857 | Avery | Sep 2005 | A1 |
| 20050197971 | Kettner et al. | Sep 2005 | A1 |
| 20050256778 | Boyd et al. | Nov 2005 | A1 |
| 20050267831 | Esary et al. | Dec 2005 | A1 |
| 20050278227 | Esary et al. | Dec 2005 | A1 |
| 20060004861 | Albanese et al. | Jan 2006 | A1 |
| 20060069585 | Springfield et al. | Mar 2006 | A1 |
| 20060241923 | Xu et al. | Oct 2006 | A1 |
| Number | Date | Country |
|---|---|---|
| WO 9960486 | Nov 1999 | WO |
| WO 0029995 | May 2000 | WO |
| WO 2005119500 | Dec 2005 | WO |
