CN113658681A - Decision tree-based drug abstinence personnel abstinence effect evaluation method - Google Patents
Decision tree-based drug abstinence personnel abstinence effect evaluation method Download PDFInfo
- Publication number
- CN113658681A CN113658681A CN202110864809.9A CN202110864809A CN113658681A CN 113658681 A CN113658681 A CN 113658681A CN 202110864809 A CN202110864809 A CN 202110864809A CN 113658681 A CN113658681 A CN 113658681A
- Authority
- CN
- China
- Prior art keywords
- decision tree
- sample
- drug
- standard deviation
- data
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a decision tree-based method for evaluating the abstinence effect of drug addicts, and belongs to the technical field of machine learning. The method comprises the following technical steps: s1: selecting an objective function; s2: selecting characteristics; s3: training process: s4: and (6) evaluating the process. The drug-abstaining personnel abstinence effect evaluation model obtained by the method can be used for carrying out regular abstinence effect evaluation on drug-abstaining personnel, the evaluation only needs to extract data from an information system database, the evaluation process is simple and low in cost, additional human subjective judgment is not needed, the accuracy is high, the output index is easy to understand and grasp, the flexibility is high, the model can adapt to different systems of various regions and huge differences brought by different technical equipment, the adaptability is strong, and when the systems change and the technical progress causes huge changes of the data, the model can be quickly adapted to the changes in a mode of retraining the model.
Description
Technical Field
The invention belongs to the technical field of machine learning, and particularly relates to a decision tree-based method for evaluating the abstinence effect of drug addicts.
Background
Although many methods for evaluating the abstinence effect of forced drug addicts are proposed at present, the problems of high operation difficulty and low reliability evaluation effectiveness generally exist in the actual operation process. In addition, the design of the existing evaluation method is based on experience, parameters cannot be changed rapidly and flexibly, and the method is difficult to adapt to environmental changes caused by new technology development and information system and related system changes.
The existing drug rehabilitation information system has a large amount of data directly related to the treatment effect, such as scoring and checking data, examination results, medical examination results, rehabilitation training data and the like; however, these data lack unified standards, the difference between various regions is huge, and each institutional change and technological progress can cause the data to change greatly, and it is difficult and not intuitive to evaluate the withdrawal effect directly from these data by means of manual analysis, and the accuracy of the evaluation result depends heavily on the experience of the evaluator.
Disclosure of Invention
Aiming at the problems, the invention provides a decision tree-based withdrawal effect evaluation method for drug addicts, which can be used for carrying out regular withdrawal effect evaluation on the drug addicts by an obtained withdrawal effect evaluation model of the drug addicts through target function selection, characteristic selection, a training process and an evaluation process, wherein the evaluation only needs to extract data from an information system database, the evaluation process is simple and low in cost, no additional artificial subjective judgment is needed, the accuracy is high, the output index is easy to understand and grasp, the flexibility is high, the method can adapt to the huge differences brought by different systems and different technical equipment in various regions, the adaptability is strong, and when the system changes and the technical progress causes the huge changes of the data, the method can adapt to the changes quickly in a mode of retraining the model.
The invention is realized by the following technical scheme:
a decision tree-based method for evaluating the withdrawal effect of drug addicts comprises the following steps:
s1: selecting an objective function: selecting one dimension YD from the multi-dimensional data of the drug-dropping personnel as a target function;
s2: selecting characteristics: selecting a set of features FD from the multi-dimensional data of the drug addict;
s3: training process: establishing a training data set TrainSet according to the objective function YD and the characteristic FD, training a decision tree regression model DTM, calculating parameters LNSTD and LNMEAN of each leaf node in the model DTM, saving the decision tree regression model DTM, the overall mean GMEAN, the overall standard deviation GSTD, the sample standard deviation LNSTD and the sample mean LNMEAN, and finishing the training process;
s4: an evaluation process; loading a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN which are saved in the training process, predicting a target function YD value of a person to be evaluated according to the model DTM by using a decision tree regression algorithm, obtaining a hit leaf node of the decision tree regression model DTM to calculate LSS, calculating GSS according to the target function YD value of the person to be evaluated, the overall mean value GMEAN and the overall standard deviation GSTD, and outputting the LSS and the GSS as an evaluation result.
Further, in step S3, the trained decision tree regression model DTM is obtained by extracting samples whose month is equal to mi from the data set TrainSet, and putting the samples into the subset ModelTrainSet for training the decision tree regression model DTM, that is, the data in month mi is extracted to train the decision tree regression model DTM, and mi is equal to 12 or the median of month.
Further, in step S3, in the process of training the decision tree regression model DTM by the subset ModelTrainSet, the minimum sample number of the leaf nodes > MNS is controlled, where 10 ≦ MNS < the total number of the ModelTrainSet samples of the subset or the total number of the leaf nodes.
Further, in step S3, the parameters LNSTD and LNMEAN of each leaf node in the decision tree regression model DTM are to put all leaf nodes in the obtained decision tree regression model DTM into a unified leaf node array lnodes, the number of leaf nodes is lnsize, which is equal to the length of lnodes, the standard deviation LNSTD array and the mean LNMEAN array of all dataset transet samples hitting the leaf nodes are calculated, both the sample standard deviation LNSTD and the sample mean LNMEAN are two-dimensional arrays, the first dimension represents month and the length is 36, the second dimension represents nodes, the length is lnsize, the value of lnd [ m ] [ i ] is the standard deviation of label of the mth month sample hitting the ith leaf node, and the value of LNMEAN [ m ] [ i ] is the average of label of the mth month sample hitting the ith leaf node.
Further, in step S3, the specific calculation method of the sample standard deviation LNSTD and the sample mean LNMEAN is as follows:
s301: establishing a set array TSS, wherein the set array TSS is a two-dimensional array, the first dimension represents a month, the length is 36, the second dimension represents a node, the length is the number lnsize of leaf nodes, and all elements of the set array TSS are initialized to be an empty set;
s302: enumerating each sample x in a data set TrainSet set, predicting a predicted value py of x.features by using a decision tree regression algorithm, ignoring the predicted value py, taking a subscript lni of a leaf node of a decision tree hit in the prediction process in a leaf node array lnodes, and adding the sample x into a subset TSS [ x.month ] [ lni ];
s303: enumerating each element TSS [ m ] [ j ] of the set array TSS, wherein TSS [ m ] [ j ] is a subset of a sample, calculating the mean value and standard deviation of elements label of the subset, and storing the mean value and standard deviation into LNSTD [ m ] [ i ] and LNMEAN [ m ] [ i ];
s304: establishing a one-dimensional set array GTSS with the length of 36, initializing all elements into an empty set, listing each sample x in a data set TrainSet set, and adding x into a subset GTSS [ x.month ];
s305: enumerating each element GTSS [ m ] of the one-dimensional set array GTSS, wherein GTSS [ m ] is a subset of a sample, calculating the mean value and standard deviation of label of all samples of the subset, and storing the mean value and standard deviation into arrays GMEAN [ m ] and GSTD [ m ], wherein GMEAN and GSTD are one-dimensional arrays and represent the mean value and standard deviation of the whole, and the subscript m represents a month.
Further, in step S3, the data set TrainSet is a sample set, each sample corresponds to data of one person in the multi-dimensional drug rehabilitation data, and each sample has three columns: according to the method, the values of an objective function YD are used as label, data of selected characteristics FD are extracted from multi-dimensional drug-dropping data to construct characteristic vectors featuress, and drug-dropping time is extracted from the multi-dimensional drug-dropping data to serve as mouth, and months are taken as units.
Further, in step S4, the specific evaluation process is:
s401: a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN obtained from a storage medium loading training process;
s402: extracting feature vectors featurees of the evaluated person by using the same method as a data set TrainSet sample featureseries, predicting an objective function YD attribute value of the featurees by using a decision tree regression algorithm according to a decision tree regression model DTM, neglecting a predicted value, obtaining subscripts lni of leaf nodes of the featurees hitting the decision tree regression model DTM, calculating the drug-dropping time month of the evaluated person, and calculating a parameter LSS (YD-LNMEAN [ m ] [ lni ])/LNSTD [ m ] [ lni ];
s403: calculating GSS ═ (YD-GMEAN [ m ])/GSTD [ m ];
s404: outputting the evaluation results LSS and GSS and the time-varying trends of the LSS and GSS indexes as visual explanations of the withdrawal effect index YD of the evaluated person;
GSS & gt 0 represents that the withdrawal effect of the evaluated person is better than the overall average level, and GSS & lt 0 represents that the withdrawal effect of the evaluated person is worse than the overall average level;
when the LSS is more than 0, the current withdrawal effect of the evaluated person is superior to the average value of similar drug-relief persons,
-1 < LSS < 1, indicating that the deviation of the mean values of the person to be evaluated and the person with similar drug addiction falls within 1 standard deviation, and labeling the withdrawal effect as "normal",
LSS < -1 > indicates that the withdrawal effect of the drug-addict is lower than the average value of similar drug-addicts and exceeds 1 standard deviation, and the withdrawal effect is marked as 'poor',
LSS >1 indicates that the withdrawal effect of the drug-addict is higher than the average value of similar drug-addicts by more than 1 standard deviation, and the withdrawal effect is marked as 'excellent';
when the evaluation results of the LSS and the GSS are different, the evaluation result of the LSS is used as a standard.
Further, in step S1, the objective function YD is any one of a cumulative award penalty, a monthly award penalty, an examination score, a medical examination result, and a rehabilitation training score.
Further, in step S2, the characteristic FD is any one of gender, age, drug type and culture degree.
Compared with the prior art, the invention has the advantages and beneficial effects that:
1. the method overcomes the defects of the existing method for evaluating the abstinence effect of the forced abstinence personnel, utilizes the data directly related to the abstinence effect in the abstinence information system to automatically extract the data from the database to construct a training set, and uses a decision tree regression algorithm to train a forced abstinence personnel abstinence effect evaluation model based on the abstinence historical data, the obtained model can carry out regular abstinence effect evaluation on the abstinence personnel, the evaluation only needs to extract the data from the information system database, no additional artificial subjective judgment needs to be added, the method is simple and easy to operate, the accuracy is high, and the output index is easy to understand and grasp.
2. The method of the invention establishes a withdrawal effect evaluation model completely based on data, eliminates human subjective factors, realizes that the model is updated at any time by constructing a data set and retraining, can quickly and flexibly adapt to the change of environment, and can also adapt to the huge difference of the technical and institutional environments of different regions.
3. The method has the advantages of simple evaluation process, low cost, easy operation and easy understanding of evaluation results; the LSS index takes the average value and the standard deviation of similar drug addicts as comparison standards, and considers the differences of sex, cultural degree and the like of the evaluated person, so that the evaluation result is more reasonable; the flexibility is high, the method can adapt to the huge difference brought by different systems and different technical equipment of each region, although the original data has huge difference, the evaluation results LSS and GSS have the same value range and similar numerical value meaning, and the method is easy to popularize; the adaptability is strong, and when the system is changed and the data is changed greatly due to the technical progress, the change can be quickly adapted in a mode of retraining the model.
Drawings
Fig. 1 is a flowchart of a training process in embodiment 1 of the present invention.
FIG. 2 is a flowchart of an evaluation process in embodiment 1 of the present invention.
Detailed Description
The present invention is further illustrated by the following examples, which are provided only for illustrating the present invention and are not intended to limit the scope of the present invention.
Example 1
A decision tree-based method for evaluating the withdrawal effect of drug addicts comprises the following steps:
s1: selecting an objective function: selecting one dimension YD from multidimensional data of drug addicts as a target function, wherein YD is a continuous real number type and is a quantitative index directly related to the abstinence effect, and selecting any one of accumulated award penalty points, monthly award penalty points, examination scores, medical examination results and rehabilitation training scores;
s2: selecting characteristics: selecting a group of characteristic FD from the multidimensional data of the drug-dropping personnel, selecting the static attribute of the drug-dropping personnel as the characteristic, namely that the attribute values are not changed in the whole drug-dropping process, and selecting any one of gender, age, drug types and cultural degree;
s3: training process: establishing a training data set TrainSet according to the objective function YD and the characteristic FD, training a decision tree regression model DTM, calculating parameters LNSTD and LNMEAN of each leaf node in the model DTM, saving the decision tree regression model DTM, the overall mean GMEAN, the overall standard deviation GSTD, the sample standard deviation LNSTD and the sample mean LNMEAN, and finishing the training process as shown in a flow chart of a training process in figure 1;
the data set TrainSet is a sample set, each sample corresponds to data of one person in the multi-dimensional drug rehabilitation data, and each sample has three columns: according to the method, month, label and features are used, the value of an objective function YD is used as label, data of a selected feature FD is extracted from multi-dimensional drug-dropping data to construct feature vectors, the drug-dropping time is extracted from the multi-dimensional drug-dropping data to serve as mouth, and a month is used as a unit;
the training decision tree regression model DTM is that samples with month equal to mi are extracted from a data set TrainSet, a subset ModelTrainSet is put into the samples for training the decision tree regression model DTM, namely the data in the month mi are extracted to train the decision tree regression model DTM, and mi is the middle value of month or mi is 12; in the process of training a decision tree regression model DTM by the subset ModelTrainSet, controlling the minimum sample number of leaf nodes to be more than MNS, wherein MNS is more than or equal to 10 and less than the total number of the ModelTrainSet samples or the total number of the leaf nodes;
calculating parameters LNSTD and LNMEAN of each leaf node in the decision tree regression model DTM, namely putting all leaf nodes in the obtained decision tree regression model DTM into a unified leaf node array lnodes, wherein the number of the leaf nodes is lnize which is equal to the length of the lnodes, calculating a standard deviation LNSTD array and a mean LNMEAN array of all dataset TransSet samples hitting the leaf nodes, wherein the sample standard deviation LNSTD and the sample mean LNMEAN are two-dimensional arrays, the first dimension represents months and the length is 36, the second dimension represents nodes, the length is lnize, the value of LNSTD [ m ] [ i ] is the standard deviation of label of an mth month sample hitting the ith leaf node, and the value of LNMEAN [ m ] [ i ] is the mean of label of the mth month sample hitting the ith leaf node;
the specific calculation method of the sample standard deviation LNSTD and the sample mean LNMEAN comprises the following steps:
s301: establishing a set array TSS, wherein the set array TSS is a two-dimensional array, the first dimension represents a month, the length is 36, the second dimension represents a node, the length is the number lnsize of leaf nodes, and all elements of the set array TSS are initialized to be an empty set;
s302: enumerating each sample x in a data set TrainSet set, predicting a predicted value py of x.features by using a decision tree regression algorithm, ignoring the predicted value py, taking a subscript lni of a leaf node of a decision tree hit in the prediction process in a leaf node array lnodes, and adding the sample x into a subset TSS [ x.month ] [ lni ];
s303: enumerating each element TSS [ m ] [ j ] of the set array TSS, wherein TSS [ m ] [ j ] is a subset of a sample, calculating the mean value and standard deviation of elements label of the subset, and storing the mean value and standard deviation into LNSTD [ m ] [ i ] and LNMEAN [ m ] [ i ];
s304: establishing a one-dimensional set array GTSS with the length of 36, initializing all elements into an empty set, listing each sample x in a data set TrainSet set, and adding x into a subset GTSS [ x.month ];
s305: enumerating each element GTSS [ m ] of a one-dimensional set array GTSS, wherein GTSS [ m ] is a subset of a sample, calculating the mean value and standard deviation of label of all samples of the subset, storing the mean value and standard deviation into arrays GMEAN [ m ] and GSTD [ m ], wherein GMEAN and GSTD are one-dimensional arrays and represent the mean value and standard deviation of the whole, and subscript m represents a month;
s4: an evaluation process; loading a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN stored in a training process, predicting a target function YD value of a person to be evaluated according to the model DTM by using a decision tree regression algorithm, obtaining leaf nodes of the hit decision tree regression model DTM to calculate LSS, calculating GSS according to the target function YD value of the person to be evaluated, the overall mean value GMEAN and the overall standard deviation GSTD, and outputting LSS and GSS as an evaluation result, wherein an evaluation process flow chart is shown in FIG. 2;
the specific evaluation process is as follows:
s401: a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN obtained from a storage medium loading training process;
s402: extracting feature vectors featurees of the evaluated person by using the same method as a data set TrainSet sample featureseries, predicting an objective function YD attribute value of the featurees by using a decision tree regression algorithm according to a decision tree regression model DTM, neglecting a predicted value, obtaining subscripts lni of leaf nodes of the featurees hitting the decision tree regression model DTM, calculating the drug-dropping time month of the evaluated person, and calculating a parameter LSS (YD-LNMEAN [ m ] [ lni ])/LNSTD [ m ] [ lni ];
s403: calculating GSS ═ (YD-GMEAN [ m ])/GSTD [ m ];
s404: outputting the evaluation results LSS and GSS and the time-varying trends of the LSS and GSS indexes as visual explanations of the withdrawal effect index YD of the evaluated person;
GSS & gt 0 represents that the withdrawal effect of the evaluated person is better than the overall average level, and GSS & lt 0 represents that the withdrawal effect of the evaluated person is worse than the overall average level;
when the LSS is more than 0, the current withdrawal effect of the evaluated person is superior to the average value of similar drug-relief persons,
-1 < LSS < 1, indicating that the deviation of the mean values of the person to be evaluated and the person with similar drug addiction falls within 1 standard deviation, and labeling the withdrawal effect as "normal",
LSS < -1 > indicates that the withdrawal effect of the drug-addict is lower than the average value of similar drug-addicts and exceeds 1 standard deviation, and the withdrawal effect is marked as 'poor',
LSS >1 indicates that the withdrawal effect of the drug-addict is higher than the average value of similar drug-addicts by more than 1 standard deviation, and the withdrawal effect is marked as 'excellent';
when the evaluation results of the LSS and the GSS are different, the evaluation result of the LSS is used as a standard.
Example 2
According to the method of the embodiment 1 of the invention, the method is used for testing at a certain drug rehabilitation bureau, extracting 762 dimensional data such as basic information of 13126 drug rehabilitation personnel who have left the country since 2016-09-01, SCL90 scale test results, score assessment and the like in a drug rehabilitation law enforcement platform database, constructing a training data set TrainSet after data cleaning, error deletion and too low quality data, selecting a cumulative prize penalty as YD, mi for 12, training a model for evaluating the withdrawal effect, and then evaluating the withdrawal effect of 6971 drug rehabilitation personnel on the book.
64836 evaluation results (one for each month for each drug-addict) were obtained, with 92.7% of the results showing a consistent LSS and GSS evaluation.
Of these, 7.3% of the results (528 people involved) scored lower than the global mean, but LSS >1, i.e. the abstinence effect was "excellent"; to verify the 7.3% accuracy of the results, 20 results were randomly selected from the interval and manually evaluated by experts, of which 18 were excellent and 2 were normal, i.e., the LSS evaluation accuracy of the data in the interval was 90%.
Therefore, the invention considers and analyzes the combination of the LSS and the GSS, carries out comprehensive evaluation, not only improves the evaluation efficiency and confirms the accuracy of most data, but also has higher accuracy, so that about 470 drug addicts which have no outstanding score but actually have good performance obtain more fair evaluation.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and the like that are within the spirit and principle of the present invention are included in the present invention.
Claims (9)
1. A decision tree-based method for evaluating the withdrawal effect of drug addicts is characterized by comprising the following steps:
s1: selecting an objective function: selecting one dimension YD from the multi-dimensional data of the drug-dropping personnel as a target function;
s2: selecting characteristics: selecting a set of features FD from the multi-dimensional data of the drug addict;
s3: training process: establishing a training data set TrainSet according to the objective function YD and the characteristic FD, training a decision tree regression model DTM, calculating parameters LNSTD and LNMEAN of each leaf node in the model DTM, saving the decision tree regression model DTM, the overall mean GMEAN, the overall standard deviation GSTD, the sample standard deviation LNSTD and the sample mean LNMEAN, and finishing the training process;
s4: an evaluation process; loading a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN which are saved in the training process, predicting a target function YD value of a person to be evaluated according to the model DTM by using a decision tree regression algorithm, obtaining a hit leaf node of the decision tree regression model DTM to calculate LSS, calculating GSS according to the target function YD value of the person to be evaluated, the overall mean value GMEAN and the overall standard deviation GSTD, and outputting the LSS and the GSS as an evaluation result.
2. The method of claim 1, wherein in step S3, the decision tree regression model DTM is trained by extracting samples with month equal to mi from the data set TrainSet, and putting the subset ModelTrainSet in the training decision tree regression model DTM, i.e. extracting the data in month mi to train the decision tree regression model DTM, wherein mi is the middle value of month or is 12.
3. The method of claim 2, wherein in step S3, the minimum number of samples of leaf nodes > MNS is controlled during the subset ModelTrainSet training decision tree regression model DTM, wherein 10 < MNS < the total number of the ModelTrainSet samples or the total number of leaf nodes.
4. The method of claim 1 wherein the method comprises the step of evaluating the abstinence efficiency of the drug addicts, in step S3, the parameters LNSTD and LNMEAN of each leaf node in the decision tree regression model DTM are to put all leaf nodes in the obtained decision tree regression model DTM into a uniform leaf node array lnodes, the number of leaf nodes is lnsize, which is equal to the length of lnodes, the standard deviation LNSTD array and the mean LNMEAN of all dataset transet samples hitting the leaf nodes are calculated, both the sample standard deviation LNSTD and the sample mean LNMEAN are two-dimensional arrays, the first dimension represents month and the length is 36, the second dimension represents nodes, the length is lnsize, the value of lnd [ m ] [ i ] is the standard deviation of label of the mth month sample hitting the ith leaf node, and the value of LNMEAN [ m ] [ i ] is the average of label of the mth month sample hitting the ith leaf node.
5. The method for evaluating the abstinence effect of drug addicts based on decision trees according to claim 4, wherein in step S3, the specific calculation methods of the sample standard deviation LNSTD and the sample mean LNMEAN are as follows:
s301: establishing a set array TSS, wherein the set array TSS is a two-dimensional array, the first dimension represents a month, the length is 36, the second dimension represents a node, the length is the number lnsize of leaf nodes, and all elements of the set array TSS are initialized to be an empty set;
s302: enumerating each sample x in a data set TrainSet set, predicting a predicted value py of x.features by using a decision tree regression algorithm, ignoring the predicted value py, taking a subscript lni of a leaf node of a decision tree hit in the prediction process in a leaf node array lnodes, and adding the sample x into a subset TSS [ x.month ] [ lni ];
s303: enumerating each element TSS [ m ] [ j ] of the set array TSS, wherein TSS [ m ] [ j ] is a subset of a sample, calculating the mean value and standard deviation of elements label of the subset, and storing the mean value and standard deviation into LNSTD [ m ] [ i ] and LNMEAN [ m ] [ i ];
s304: establishing a one-dimensional set array GTSS with the length of 36, initializing all elements into an empty set, listing each sample x in a data set TrainSet set, and adding x into a subset GTSS [ x.month ];
s305: enumerating each element GTSS [ m ] of the one-dimensional set array GTSS, wherein GTSS [ m ] is a subset of a sample, calculating the mean value and standard deviation of label of all samples of the subset, and storing the mean value and standard deviation into arrays GMEAN [ m ] and GSTD [ m ], wherein GMEAN and GSTD are one-dimensional arrays and represent the mean value and standard deviation of the whole, and the subscript m represents a month.
6. The method of claim 1, wherein in step S3, the data set TrainSet is a set of samples, each sample corresponding to data of one person in the multi-dimensional detoxification data, and each sample has three columns: according to the method, the values of an objective function YD are used as label, data of selected characteristics FD are extracted from multi-dimensional drug-dropping data to construct characteristic vectors featuress, and drug-dropping time is extracted from the multi-dimensional drug-dropping data to serve as mouth, and months are taken as units.
7. The method for evaluating the abstinence effect of drug addicts based on decision trees as claimed in claim 1, wherein in step S4, the specific evaluation process is as follows:
s401: a decision tree regression model DTM, an overall mean value GMEAN, an overall standard deviation GSTD, a sample standard deviation LNSTD and a sample mean value LNMEAN obtained from a storage medium loading training process;
s402: extracting feature vectors featurees of the evaluated person by using the same method as a data set TrainSet sample featureseries, predicting an objective function YD attribute value of the featurees by using a decision tree regression algorithm according to a decision tree regression model DTM, neglecting a predicted value, obtaining subscripts lni of leaf nodes of the featurees hitting the decision tree regression model DTM, calculating the drug-dropping time month of the evaluated person, and calculating a parameter LSS (YD-LNMEAN [ m ] [ lni ])/LNSTD [ m ] [ lni ];
s403: calculating GSS ═ (YD-GMEAN [ m ])/GSTD [ m ];
s404: outputting the evaluation results LSS and GSS and the time-varying trends of the LSS and GSS indexes as visual explanations of the withdrawal effect index YD of the evaluated person;
GSS & gt 0 represents that the withdrawal effect of the evaluated person is better than the overall average level, and GSS & lt 0 represents that the withdrawal effect of the evaluated person is worse than the overall average level;
when the LSS is more than 0, the current withdrawal effect of the evaluated person is superior to the average value of similar drug-relief persons,
-1 < LSS < 1, indicating that the deviation of the mean values of the person to be evaluated and the person with similar drug addiction falls within 1 standard deviation, and labeling the withdrawal effect as "normal",
LSS < -1 > indicates that the withdrawal effect of the drug-addict is lower than the average value of similar drug-addicts and exceeds 1 standard deviation, and the withdrawal effect is marked as 'poor',
LSS >1 indicates that the withdrawal effect of the drug-addict is higher than the average value of similar drug-addicts by more than 1 standard deviation, and the withdrawal effect is marked as 'excellent';
when the evaluation results of the LSS and the GSS are different, the evaluation result of the LSS is used as a standard.
8. The method of claim 1, wherein the objective function YD is any one of a cumulative award penalty, a monthly award penalty, an examination score, a medical examination result, and a rehabilitation training score in step S1.
9. The method of claim 1, wherein the characteristic FD is any one of gender, age, drug type and cultural degree in step S2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110864809.9A CN113658681A (en) | 2021-07-29 | 2021-07-29 | Decision tree-based drug abstinence personnel abstinence effect evaluation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110864809.9A CN113658681A (en) | 2021-07-29 | 2021-07-29 | Decision tree-based drug abstinence personnel abstinence effect evaluation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113658681A true CN113658681A (en) | 2021-11-16 |
Family
ID=78490851
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110864809.9A Pending CN113658681A (en) | 2021-07-29 | 2021-07-29 | Decision tree-based drug abstinence personnel abstinence effect evaluation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113658681A (en) |
-
2021
- 2021-07-29 CN CN202110864809.9A patent/CN113658681A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109857835B (en) | Self-adaptive network security knowledge evaluation method based on cognitive diagnosis theory | |
| CN107545245A (en) | A kind of age estimation method and equipment | |
| CN109002492B (en) | Performance point prediction method based on LightGBM | |
| CN108304887A (en) | Naive Bayesian data processing system and method based on the synthesis of minority class sample | |
| CN107274020A (en) | A kind of learner's subject based on collaborative filtering thought always surveys result prediction system and method | |
| CN107610009B (en) | Trinity enrollment probability prediction method based on neural network | |
| CN109635083A (en) | It is a kind of for search for TED speech in topic formula inquiry document retrieval method | |
| CN106529110A (en) | Classification method and equipment of user data | |
| CN105956798A (en) | Sparse random forest-based method for assessing running state of distribution network device | |
| CN114429212A (en) | Intelligent learning knowledge ability tracking method, electronic device and storage medium | |
| CN109740672A (en) | Multi-streaming feature is apart from emerging system and fusion method | |
| CN114971425A (en) | Database information monitoring method, device, equipment and storage medium | |
| CN103150476B (en) | A kind of system efficiency evaluation method based on data station field | |
| CN107679174A (en) | Construction method, device and the server of Knowledge Organization System | |
| CN111461529B (en) | Quality collaborative evaluation method and system for fresh aquatic products | |
| CN112733340A (en) | Well selection method and equipment for modifying candidate well based on data-driven reservoir | |
| CN113658681A (en) | Decision tree-based drug abstinence personnel abstinence effect evaluation method | |
| CN107729510A (en) | Information interacting method, information interactive terminal and storage medium | |
| CN113657726B (en) | Personnel risk analysis method based on random forest | |
| CN110210768A (en) | A kind of classic poetry item difficulty appraisal procedure and system | |
| CN113889274B (en) | Method and device for constructing risk prediction model of autism spectrum disorder | |
| CN113658680B (en) | Evaluation method for drug-dropping effect of drug-dropping personnel based on random forest | |
| CN109493975A (en) | Chronic disease recurrence prediction method, apparatus and computer equipment based on xgboost model | |
| CN113919983A (en) | Test question portrait method, device, electronic equipment and storage medium | |
| Gupta | Managerial judgment and forecast combination: An experimental study |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |