TWI418995B - The Method of Establishing Quantified Sequence Tree and Its Computer Program - Google Patents

Description

Method for establishing quantized sequence tree and computer program thereof

A method for establishing a sequence tree, in particular, a method for establishing a sequence tree of quantization.

Please refer to FIG. 1. In the prior art, the detecting personnel use detecting instruments (such as physiological signal detectors: sphygmomanometers, blood glucose meters, pulse counters, etc.; circuit signal detectors: galvanometers, voltmeters, digits) A signal detector, a frequency detector, measures a measured object (home patient, digital circuit, analog circuit, etc.) to obtain a time series data and then introduces it into a two-dimensional coordinate axis, the two-dimensional coordinate axis One axis is the time axis and the other axis is the variation range of the time series data. The detecting personnel defines values in at least one region specified by the variation range, and thus the region defining values divide the two-dimensional coordinate axes into a plurality of state intervals, and then the time series data is cut into equal data segments to form a plurality of data segments, and Calculate the average of each data segment. The detecting instrument will judge the order of the status interval and the time segment of each average value, and sequentially convert each data segment into a string of data for transmission to a special string interpretation device, and interpret the string data to facilitate the interpretation. Professionals conduct follow-up data analysis. For example, establishing a sequence tree, establishing a decision tree, or judging the actual state change of the measured object by other means.

However, the basis for establishing string data is generated by the flat value and order of each data segment. If the data of a data segment changes quite large, such as the data segment curve is continuously shifted back and forth between the a and c regions. The resulting flat The average value is likely to correspond to the b-th area, and the generated string data is the incorrect string data. The string data read by the string interpretation device is the wrong value, and it is impossible for the professional to accurately and correctly understand the actual state change of the measured object, that is, the result analyzed by the string data is an error.

Therefore, how to convert time series data into correct string data, and effectively use string data to analyze the state change of the measured object is a problem that should be considered before the current factory.

In view of the above, the problem to be solved by the present invention is to provide a data classification method for converting correct string data from time series data and analyzing the state of the measured object by using string data.

In order to solve the above problem of data analysis, the technical means provided by the present invention discloses a method for establishing a quantitative sequence tree, which is to obtain at least one string of data, the string data includes at least one data function, and each data function includes one The interval coding and the one-time segment sequentially convert the data function into at least one node, and import the node into a tree structure according to the order of the data function. When the parent node of the import node has at least one child node, and the import node and the child node of the parent node are the same interval code and time segment, the statistics value of the import node is added to the statistics of any child node. . Conversely, the import node is combined with any of the above child nodes as a combined node to serve as a child node of the parent node. When all string data has been processed and imported into the tree A structure in which a joint node having a data dependency relationship or a child node having the same parent node is merged according to a data dependent condition.

The string data disclosed in the present invention is obtained by a time series data classification method. The method maps a time series data to a two-dimensional coordinate axis, calculates at least one region defining value of the data variation range of the time series data and the data variation range, and divides the two-dimensional coordinate axis into a plurality of state intervals according to the regional definition value, and calculates the time. The sequence data and the region-defined value are at least one time tangent point, and the time-series data is divided into a plurality of data segments according to the time-cut point, and all the data segments are coded according to the interval code of the time segment and the state interval of each data segment. The above string data.

The method proposed by the present invention can also be presented in the form of a computer program product or a recording medium. When the electronic device capable of reading a computer program or a recording medium is loaded into the computer program or the recording medium, the same can be achieved. The method solves the same problem and achieves the same effect.

The achievable effects of the present invention include:

First, the data segment cut out from the time series data, the calculated data average and the data segment curve, corresponding to the same state interval and time segment. There will be no situation where the data segment curve changes greatly during a time period, so the converted string data is more accurate.

Secondly, converting the string data into nodes to establish a quantized sequence tree is beneficial to simplify the complexity of the sequence data, and at the same time, by quantifying the sequence tree, a more accurate decision tree can be established, which is useful for knowing the actual state of the measured object. Variety.

In order to further understand the object, structural features and functions of the present invention, the related embodiments and drawings are described in detail as follows: Please refer to FIG. 2, FIG. 3 and FIG. 4 simultaneously, FIG. 2 is the time of the embodiment of the present invention. FIG. 3 is a schematic diagram of a state interval division according to an embodiment of the present invention, and FIG. 4 is a schematic diagram of data segment division according to an embodiment of the present invention. The method is described as follows: the detection personnel use detection instruments (such as physiological signal detectors: blood pressure monitors, blood glucose meters, pulse counters, etc.; circuit signal detectors: galvanometers, voltmeters, digital signal detectors) And a frequency detecting meter for measuring a measured object (home patient, digital circuit, analog circuit, etc.) to obtain a time series data 201, and mapping a time series data 201 to a two-dimensional coordinate axis (step S110) . One axis of the two-dimensional coordinate axis is the time axis, and the other axis is used to display the data variation range 202 of the time series data 201.

Obtaining at least one region defining value of the data variation range of the time series data and the data variation range (step S120). In this step, the time series data changes are detected for a period of time to obtain the maximum and minimum values of the time series data. The upper limit of the data variation range is the maximum value of the time series data, and the lower limit value of the data variation range is the minimum value of the time series data.

The detection personnel assign at least one regional definition value to the data variation range. The regional definition value is provided by the expert system or professional, such as to track the patient's physiological signal, which is recommended by the physician or caregiver. For example, to track the signal changes of the circuit, this value is provided by the designer.

The two-dimensional coordinate axis is divided into a plurality of state intervals according to at least one region defining value, and each state interval corresponds to a different one-section encoding (step S130). Here, the value of each region is defined as the upper or lower limit of the state interval, and each state interval represents a data change function. When the time series data is a physiological signal, the area a represents a poor state. The b-zone represents a medium state, and the c-zone represents a good state, that is, the value in the c-zone represents a good condition; or conversely, the c-zone represents a poor state, the b-zone represents a moderate state, and the a-region represents a The condition is very good, that is, the value in the a area indicates that the patient is in good condition. The most important thing is to judge the migration of time series data in each state interval. The divided state intervals can be the same area or different areas, mainly based on the needs of professionals.

At least one time tangent point 203 of the time series data 201 and the at least one region defined value is calculated (step S140). As shown in FIG. 3, the intersection of the time series data 201 and the boundary of each state area (ie, the defined value of each area) is found, and the intersection is the time tangent point 203.

The time series data 201 is divided into a plurality of data segments based on at least one time tangent point 203 (step S150). As shown in Fig. 4, the time unit corresponding to the time tangent point is found from the time axis of the two-dimensional coordinate axis. The time series data is divided according to the time segments formed by the time units to form a plurality of data segments. Here, the data section of the same time zone does not span two More than one status interval will only be in a single status interval.

According to the interval encoding of the time segment of each data segment and the state interval, the data segments are encoded as a string of data (step S160). When each data section is encoded to form a data function, each data function includes a section of the status section of the corresponding data section and a time section of the data area, and the value stored in the data function is The average value of the data section. All data functions are coded as the above-mentioned string data according to the chronological order of each data section. In the case of Fig. 4, the formed string data is (b, 18) (a, 35) (b, 6) (c, 32) (b, 17) (c, 20).

Please refer to FIG. 5, which is a flowchart of a method for establishing a quantized sequence tree according to an embodiment of the present invention. This method is explained as follows: At least one string of data is obtained (step S210). Each string data includes at least one data function, and each data function includes an interval code, a time segment and a data category. Referring to FIG. 6, the following plurality of string data shown in FIG. 6 will be used as input data.

The data function of each string data is sequentially converted into a node (step S220), and each node includes an interval code, a time segment and an appearance statistical value of the data function to which it belongs.

The nodes are introduced into a tree structure in accordance with the order of the respective data functions (step S230). Referring to FIG. 7, the tree structure referenced in this embodiment is referred to herein as a tree structure of a quantization sequence tree (QFS-Tree). The root of this quantized sequence tree, whose child nodes are the data categories of the string data. When a node is imported into a tree structure, the node is constructed according to the data category to which it belongs. Form more than one branch.

Please refer to FIG. 8 and FIG. 7 simultaneously. FIG. 8 is a schematic diagram of a joint node according to an embodiment of the present invention. When a parent node to which an import node belongs has at least one child node, it is determined whether the child node of the import node and the parent node have the same interval code and time segment. The third string data shown in FIG. 6 will be described as the imported string data.

First, it is judged whether or not the section code of the import node is the same as the section code of any of the child nodes (step S240).

When it is determined that the interval is encoded, the import node forms a child node of the parent node (step S241), that is, a child node is added below the node data 301.

On the other hand, if it is determined that the same section code is used, it is judged whether or not the time zone of the import node is the same as the time zone of any of the child nodes (step S242).

When it is determined that the different time zone is combined, the combined import node and any of the child nodes are a combined node 302 (step S243). The interval code of the combined node is the interval code of any child node, and the combined node includes the time segment of the imported node and the occurrence statistics, and the time segment of any child node and the occurrence statistics.

When it is determined that the same time zone is present, the occurrence statistics of the import node are added to the occurrence statistics of any of the child nodes (step S244). That is, as shown in the node data 301 shown in FIG. 8, the occurrence statistical value changes from 1 to 2.

It is judged whether or not each of the string data has been completely processed and introduced into the tree structure (step S250).

When it is judged that it is not processed in full, that is, step S210 is re-executed to guide Enter the next round of string information.

Please refer to FIG. 9, which is a schematic diagram of the dependent data of the quantized sequence tree of the present invention. As shown in FIG. 9, first, a data dependency condition is first obtained, and the data dependency condition is a previously established joint node, and the difference between at least two time segments is less than a specified value, and the following values are used as an explanation.

Next, there is at least one bonding node from the tree structure. As shown in FIG. 9, the node shown by the dashed box is the previously described bonding node.

Next, it is determined whether at least two time segments of the discovered bonding node meet the data dependent condition. As shown in FIG. 9, the joints shown in the dashed box contain two time zones which are adjacent values, so that the above-mentioned data dependent conditions are met. Therefore, the above-described combining node 302 is converted into the merge node 303. That is, as shown in FIG. 10, the occurrence statistics of the merge node 303 are the sum of the occurrence statistics corresponding to the time segments of the previous join node 302. On the other hand, if it is determined that at least two time segments of the discovered bonding node do not meet the data dependency condition, no node conversion behavior is performed on the bonding node 302.

Next, it is determined whether the number of child nodes of each merge node 303 is a majority. If any one of the merge nodes 303 includes multiple child nodes, whether the child nodes meet the above-mentioned data dependency condition. If so, the child nodes that meet the data dependency condition are merged into one merged child node 304. Referring to FIG. 10, after the node 302 is converted to the merge node 303, its child nodes also meet the data dependency condition, that is, the time segments of the two child nodes are adjacent values. Therefore, the two child nodes are merged into a merged child node 304, and the interval code of the merged child node 304 corresponds to the area of the two child nodes at the same time. Inter-coded, and the occurrence statistics of the merged child nodes are the sum of the occurrence statistics of the two child nodes before the merge. However, among the two child nodes that are merged, if one of the child nodes belongs to the joint node type, another child node is coupled to the child node of the joint node 303 type according to the determination mode of step S240 and step S242. The final tree structure is formed as shown in Figure 11, as a quantitative sequence tree for completing the construction work.

Generally, during the establishment of the sequence tree, the connection relationship of each data node is recorded in a header table according to more than one reference value. For the present embodiment, the file header table uses the interval coding as a reference value to record the connection relationship between the nodes having the same interval coding and other nodes, and the nodes having the same interval coding are linked according to the data category. At the same time, the file header table records the order of the encoding of each interval as a sequence of capturing the sequence of the sequence from the data categories of the quantized sequence tree.

As shown in FIG. 12, it is assumed here that the sequence extraction condition is to perform sequence extraction on the quantized sequence tree in the order of the interval code c, the interval code b, and the interval code a. If the quantized sequence tree has more than two branches, the sequence of each branch is separately extracted, and the sequence string extracted by the branch is classified according to the data category to which each branch belongs. As shown in Fig. 12, category 1 has two branches, and the strings extracted by each of the two branches are (A, 2) (B: 5~6) (C: 1~2): 3, and (B). :5)(C:5): 1.

FIG. 13 is a schematic diagram of sequence string deletion according to an embodiment of the present invention. When the sequence string is extracted, the result is as shown in FIG. 13 . Here, the description is Conducive to the subsequent description, the sequence string shown in FIG. 13 is not the result of the above-mentioned string, but is assumed by the range in which it is related.

After that, any sequence string corresponding to all data categories is deleted at the same time. In FIG. 13, the sequence string of (B, 2~5) is deleted. Finally, after the sequence string deletion operation is completed, the remaining result sequence strings are imported into a number decision tree, as shown in FIG. 14, for the patenter to know the state value change of the measured object.

While the present invention has been described above in terms of the preferred embodiments thereof, it is not intended to limit the invention, and the equivalent of the modification and retouching of the present invention is still within the spirit and scope of the present invention. Within the scope of patent protection of the present invention.

201‧‧‧ Time Series Information

202‧‧‧Scope of data change

203‧‧‧ time cut point

301‧‧‧data node

302‧‧‧Combined nodes

303‧‧‧ merged nodes

304‧‧‧Combined child nodes

1 is a prior art string data construction diagram; FIG. 2 is a flow chart of a time series data classification method according to an embodiment of the present invention; FIG. 3 is a schematic diagram of a state interval division according to an embodiment of the present invention; FIG. 5 is a flowchart of a method for establishing a quantized sequence tree according to an embodiment of the present invention; FIG. 6 is a list of string data according to an embodiment of the present invention; FIG. 7 is a schematic diagram of node import according to an embodiment of the present invention; FIG. 9 is a schematic diagram of a dependency of a quantized sequence tree according to an embodiment of the present invention; FIG. 10 is a schematic diagram of a node merging according to an embodiment of the present invention; FIG. 12 is a schematic diagram of sequence capture according to an embodiment of the present invention; FIG. 13 is a schematic diagram of sequence string deletion according to an embodiment of the present invention; and FIG. 14 is a schematic diagram of a number decision tree according to an embodiment of the present invention.

Claims (22)

A method for establishing a sequence tree includes: obtaining at least one string of data, the at least one string of data comprising at least one data function, each data function comprising an interval code and a time segment; Converting a data function into at least one node, each node including an interval code, a time segment and an occurrence statistical value of the data function to which the data belongs; and importing the at least one node into a tree according to the order of the at least one data function Structure; when a parent node to which an import node belongs has at least one child node, it is determined whether the import node and any child node of the parent node have the same interval code and time segment; when it is judged to be the same, the import node The occurrence statistics value is added to the occurrence statistics of any one of the child nodes; when it is judged to be different, the import node is formed as a child node of the parent node; determining whether the at least one string data has been completely processed and imported into the tree structure; When it is determined that the data is not processed in full, that is, any string data is taken out from the at least one string of data, and the at least one data function is sequentially converted into This step is a little information on the node. The method for establishing a quantized sequence tree according to claim 1, wherein the step of determining whether the import node and the child node of the parent node have the same interval code and time segment further comprises the following steps: Determining whether the interval code of the imported node is the same as the interval code of the any child node; when it is determined that the interval code is different, the import node is imported to form a child node of the parent node; when it is determined that the same interval code is determined, the import node is determined Whether the time segment is the same as the time segment of any one of the child nodes; when it is determined that the time zone is different, the combined node is combined with the child node as a combined node, and the interval of the combined node is coded as any one of Interval coding of the child node, and the binding node includes a time zone and an occurrence statistical value of the import node, and a time zone and an occurrence statistical value of the any child node; and when the same time zone is determined, the import is performed The statistics of the occurrence of the node are added to the statistics of the occurrence of any of the child nodes. The method for establishing a quantized sequence tree according to claim 2, further comprising the steps of: obtaining a data dependent condition; determining whether the tree structure has at least one binding node; and determining that the existence exists, determining the at least one bonding node Whether at least two time segments meet the data dependent condition; and when determined to be consistent, the at least one combined node is converted into at least one merged node. The method for establishing a quantized sequence tree according to claim 3, wherein the time segment of the at least one merged node corresponds to at least one combined node At least two time segments, and the data dependent condition is that the difference between the at least two time segments of the at least one bonding node is less than a specified value, and the occurrence statistics of the at least one combining node are the at least one of the at least one bonding node The sum of the occurrence statistics corresponding to the two time segments. The method for establishing a quantized sequence tree according to claim 3, further comprising the steps of: determining whether a child node of the at least one binding node is a majority; and determining, when the child node of any node is a majority, determining the at least one bonding node Whether at least two child nodes meet the data dependent condition; and when determined to be consistent, the at least two child nodes of the at least one combined node are merged into one combined child node. The method for establishing a quantized sequence tree according to claim 5, wherein the data dependent condition is that the difference between the time segments of the at least two child nodes is less than a specified value. The method for establishing a quantized sequence tree according to claim 5, wherein the interval coding system of the merged child node simultaneously encodes the intervals corresponding to at least two child nodes. The method for establishing a quantized sequence tree according to claim 5, wherein the statistical value of the occurrence of the merged child node is a sum of the statistical values of the at least two child nodes. The method for establishing a quantized sequence tree according to claim 5, wherein each string data corresponds to a data category, and when the at least one node is imported into the tree structure, the at least one data category is based on the at least one data category. One The nodes are constructed to form at least two branches. The method for establishing a quantized sequence tree according to claim 9 , further comprising the steps of: obtaining a sequence of capture conditions; and determining whether the at least two branches have at least one sequence string that meets the sequence capture condition; Determining yes, obtaining the at least one sequence string and classifying according to the data categories; deleting any sequence string corresponding to the data category at the same time; and deleting at least one result sequence word after the sequence string is deleted The string is imported into a number decision tree. The method for establishing a quantized sequence tree according to claim 1, wherein the step of obtaining at least one string of data comprises: mapping a time series data to a two-dimensional coordinate axis; calculating a data variation range of the time series data and the At least one region defining value of the data variation range; dividing the two-dimensional coordinate axis into a plurality of state intervals according to the at least one region defining value, each state interval corresponding to a different interval encoding; calculating the time series data and the at least one At least one time tangent point of the region defining value; dividing the time series data into a plurality of data segments according to the at least one time tangent point; and, according to the time segment of each data segment and the region in the state interval Inter-coded to encode the data segments as the string data. A computer program product is executed by an electronic device, the electronic device executing the computer program product system for performing a quantization sequence tree establishing method, the method comprising: obtaining at least one string of data, the at least one string of data comprising at least one data a function, each data function includes an interval coding and a time segment; sequentially converting the at least one data function into at least one node, each node including an interval code and a time segment to which the data function belongs And at least one node is imported into a tree structure according to the at least one data function; and when an import node belongs to a parent node having at least one child node, determining the import node and the parent node Whether a child node has the same interval code and time segment; when it is judged to be the same, the appearance statistical value of the import node is added to the occurrence statistics of the any child node; when it is judged to be different, the import node forms the parent a child node of the node; determining whether the at least one string of data has been fully processed and imported into the tree structure; and when it is determined that the total number is not , From which at least any one of a string data extracted string data, the sequence returns to the step of the at least one data function into at least one node of information. For example, the computer program product described in claim 12 of the patent scope, wherein the judgment The step of breaking whether the import node and the child node of the parent node have the same interval code and time segment further includes the following steps: determining whether the interval code of the import node is the same as the interval code of any one of the child nodes When it is determined that the difference interval code is encoded, the import node is imported to form a child node of the parent node; when it is determined that the same interval code is determined, it is determined whether the time segment of the import node is the same as the time segment of the any child node; For the different time segment, the import node is combined with the child node, and the interval code of the joint node is the interval code of the any child node, and the joint node includes the time segment of the import node. And the statistical value, and the time segment of the any child node and the occurrence of the statistical value; and when it is determined that the same time segment, the statistical value of the appearance of the imported node is added to the statistical value of the occurrence of the any child node. The computer program product of claim 13, further comprising the steps of: obtaining a data dependency condition; determining whether the tree structure has at least one binding node; and determining that the presence is present, determining at least one of the bonding nodes Whether the two time segments meet the data dependent condition; and when it is determined to be in compliance, converting the at least one bonding node into the at least one combining node. The computer program product of claim 14, wherein the time zone of the at least one merged node corresponds to at least one time zone of the at least one combined node, and the data dependency condition is the at least one of the at least one bonded node The difference between the two time segments is less than a specified value, and the occurrence statistics of the at least one combined node are the sum of the occurrence statistics corresponding to the at least two time segments of the at least one combined node. The computer program product of claim 14, further comprising the steps of: determining whether a child node of the at least one merge node is a majority; and when there are multiple child nodes of any merge node, determining whether the merge node is any Whether at least two child nodes meet the data dependent condition; and when determined to be consistent, the at least two child nodes of the at least one combined node are merged into one combined child node. The computer program product of claim 16, wherein the data dependent condition is that the difference between the time segments of the at least two child nodes is less than a specified value. The computer program product of claim 16, wherein the interval coding of the merged child node simultaneously encodes the intervals of at least two child nodes. The computer program product of claim 16, wherein the statistical value of the merged child node is a sum of the statistical values of the at least two child nodes. For example, the computer program products mentioned in claim 16 of the patent scope, each of which The string data corresponds to a data category. When the at least one node is imported into the tree structure, the at least one node is constructed to form at least two branches according to the at least one data category. The computer program product of claim 20, further comprising the steps of: obtaining a sequence of capture conditions; determining whether the at least two branches have at least one sequence string that meets the sequence capture condition; Yes, obtaining the at least one sequence string and classifying according to the data categories; deleting any sequence string corresponding to the data category at the same time; and importing at least one result sequence string after deleting the any sequence string A number of decision trees. The computer program product of claim 12, wherein the step of obtaining at least one string of data comprises: mapping a time series data to a two-dimensional coordinate axis; calculating a data variation range of the time series data and the data change At least one region defining a value; the two-dimensional coordinate axis is divided into a plurality of state intervals according to the at least one region defining value, each state interval corresponding to a different interval encoding; calculating the time series data and the at least one region defining At least one time tangent point of the value; dividing the time series data into a plurality of points according to the at least one time tangent point a data segment; and an interval code according to a time segment of each data segment and the state segment in which the data segment is located to encode the data segment as the string data.