CN112381510A - Salary recommendation method based on machine learning - Google Patents
Salary recommendation method based on machine learning Download PDFInfo
- Publication number
- CN112381510A CN112381510A CN202011265168.7A CN202011265168A CN112381510A CN 112381510 A CN112381510 A CN 112381510A CN 202011265168 A CN202011265168 A CN 202011265168A CN 112381510 A CN112381510 A CN 112381510A
- Authority
- CN
- China
- Prior art keywords
- value
- compensation
- market
- candidate
- average value
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000010801 machine learning Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 23
- 230000008569 process Effects 0.000 claims abstract description 14
- 238000011835 investigation Methods 0.000 claims description 35
- 238000012360 testing method Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 6
- YBJHBAHKTGYVGT-ZKWXMUAHSA-N (+)-Biotin Chemical compound N1C(=O)N[C@@H]2[C@H](CCCCC(=O)O)SC[C@@H]21 YBJHBAHKTGYVGT-ZKWXMUAHSA-N 0.000 claims description 3
- 108010014172 Factor V Proteins 0.000 claims description 3
- 230000001105 regulatory effect Effects 0.000 claims description 3
- FEPMHVLSLDOMQC-UHFFFAOYSA-N virginiamycin-S1 Natural products CC1OC(=O)C(C=2C=CC=CC=2)NC(=O)C2CC(=O)CCN2C(=O)C(CC=2C=CC=CC=2)N(C)C(=O)C2CCCN2C(=O)C(CC)NC(=O)C1NC(=O)C1=NC=CC=C1O FEPMHVLSLDOMQC-UHFFFAOYSA-N 0.000 claims description 3
- 230000007115 recruitment Effects 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000011157 data evaluation Methods 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/105—Human resources
- G06Q10/1057—Benefits or employee welfare, e.g. insurance, holiday or retirement packages
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Tourism & Hospitality (AREA)
- Medical Informatics (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a machine learning-based salary recommendation method, which comprises the following steps: collecting and establishing a candidate sample library; for the ith candidate sample, reading the value of the compensation element before entry, the value of the compensation element after entry and the value of the number of the added items of the ith candidate; training the compensation model to obtain each factor influence factor so as to obtain a final compensation model; and calculating the salary of the candidate by adopting a salary model to obtain the amplitude salary and the characteristic salary of the candidate, and further finally obtaining the salary recommendation range. According to the salary recommendation method based on machine learning, provided by the invention, based on a salary model obtained by machine learning, the interference of human factors on salary is avoided; the compensation method is beneficial to different departments to reach the consistency of compensation of candidates; embodying the personal characteristics of the candidate; the operation is simple, and the full automation can be realized. Thereby improving the effectiveness and efficiency of the recruitment process.
Description
Technical Field
The invention belongs to the technical field of machine learning application, and particularly relates to a compensation recommendation method based on machine learning.
Background
The compensation measurement and calculation of the current candidate are completely based on the statistical significance of market data, and the moderate expansion is controlled according to certain random factors; or passively according to the division of the company grades, determining the candidate meeting the conditions, which cannot completely reflect the individual characteristics of the candidate, so that more uncertain factors exist in the salary customization process of the candidate, the credibility of the salary customization result is reduced, and the requirement of the current unit for accurately formulating the salary of the candidate is difficult to meet.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a salary recommendation method based on machine learning, which can effectively solve the problems.
The technical scheme adopted by the invention is as follows:
the invention provides a machine learning-based salary recommendation method, which comprises the following steps of:
step 1, collecting and establishing a candidate person sample library; the candidate sample library comprises a plurality of candidate samples, and any ith candidate sample is represented as: fi={(f(Ci),Ni,f(Si)};
Wherein:
f(Ci) Represents the pre-entry compensation element C of the ith candidateiValue f (C) ofi);
NiThe number of bonus items representing the ith candidate;
f(Si) Represents the post-entry compensation element S of the ith candidateiValue f (S) ofi);
Step 2, dividing candidate samples in a candidate sample library into a training set and a testing set; wherein the training set has m1A sample of individual candidates; test set has m2A sample of individual candidates;
step 3, for m1The individual candidate sample performs the following compensation model training process:
step 3.1, changing i to 1;
setting initial values, including: setting the 1 st market elementValue ofMarket factor 2Value ofIndex element E of academic calendariValue f (E) ofi) Adding the value added by the item QiValue of (A) f (Q)i) And a factor V of the result of investigationiValue f (V) ofi);
Step 3.2, for the ith candidate sample, reading the pre-entry compensation element C of the ith candidateiValue f (C) ofi) And the compensation element S after workiValue f (S) ofi) Number of bonus items NiAnd company subscription element MiValue f (M)i) (ii) a Wherein, company subscribes element MiValue f (M)i) Is a known fixed value;
step 3.3, adopting an adjusting algorithm to examine the result element ViValue f (V) ofi) Adjusting to obtain an element V of the investigation resultiAnd finally obtaining a survey result element ViAverage value of (2)Element V of the investigation resultiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.3.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.3.2, determine the amplitude compensation f (Y)i1) And characteristic compensation f (Y)i2) Whether the following relation is satisfied:
f(Si)-δ≤f(Yi1)≤f(Si)+δ
f(Si)-δ≤f(Yi2)≤f(Si)+δ
wherein: delta is the market impact coefficient, which is a known set value;
if yes, executing step 3.3.3; if not, executing step 3.3.4;
step 3.3.3, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marked as the significant point, then, changing f (V) by a set adjustment step size + -f (delta V)i) To obtain a new f (V)i) The step 3.3.1 is repeated, and the process is circulated continuously until the set times are reached, and the step 3.3.5 is executed;
step 3.3.4, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marking as an invalid point; then, f (V) is changed by a set adjustment step + -f (Δ V)i) To obtain a new f (V)i) The step 3.3.1 is repeated, and the process is circulated continuously until the set times are reached, and the step 3.3.5 is executed;
step 3.3.5, the distribution of the effective points and the invalid points is counted to obtain the range of the effective point concentration, so that the element V of the investigation result is obtainediThe adjustment range of (a);
Step 3.4, adopting an adjusting algorithm to the 1 st market elementValue ofIs adjusted to obtainMarket elements 1And finally obtaining the 1 st market elementAverage value of (2)Market 1 elementAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.4.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.4.2, compensate f (Y) according to the amplitude of fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied and further the 1 st market element is satisfiedValue ofAdjusting to obtain the 1 st market element meeting the relation of the step 3.3.2And finally obtainingTo market item 1Average value of (2)
Step 3.5, adopting an adjusting algorithm to the No. 2 market elementValue ofMaking adjustment to obtain the 2 nd market elementAnd finally obtaining the 2 nd market elementAverage value of (2)Market factor 2Average value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.5.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.5.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied or not, and further the 2 nd market elementValue ofAdjusting to obtain the 2 nd market element meeting the relation of the step 3.3.2And finally obtaining the 2 nd market elementAverage value of (2)
Step 3.6, adopting an adjusting algorithm to learn the index element E of the calendariValue f (E) ofi) Regulating to obtain the academic calendar index element EiAnd finally obtaining the academic index element EiAverage value of (2)Will learn the calendar index element EiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.6.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.6.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relation formula of the step 3.3.2 is satisfied or not, and further the academic index element EiValue f (E) ofi) Adjusting to obtain the academic calendar index element E meeting the relational expression of the step 3.3.2iAnd finally obtaining the academic index element EiAverage value of (2)
Step 3.7, adding value Q to the added items by adopting an adjusting algorithmiValue of (A) f (Q)i) Adjusting to obtain the added value QiAnd finally obtaining the added value Q of the added itemiAverage value of (2)Adding the added items to the value QiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.7.1, calculate the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.7.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relation formula of the step 3.3.2 is satisfied or not, and further adding the value Q to the added itemsiValue of (A) f (Q)i) Adjusting to meet the requirement of step 3.3.2Systematic added-term value added QiAnd finally obtaining the added value Q of the added itemiAverage value of (2)
Step 3.8, so far, the following parameters of the ith candidate are obtained: element V of investigation resultiAverage value of (2)Market 1 elementAverage value of (2)Market factor 2Average value of (2)Index element E of academic calendariAverage value of (2)Sum and add value added amount QiAverage value of (2)
Will be provided withIs assigned to f (V)i),Is assigned toIs assigned toIs assigned to f (E)i),Is assigned to f (Q)i);
Let i equal i + 1; if i is greater than m1Then m is1After the personal candidate samples are trained on the compensation model, executing step 4; otherwise, returning to the step 3.2 for circular execution;
a1and b1Are respectively the 1 st market elementAnd the 1 st and 2 nd influence factors of (c), and a is set1>0,b1>0;
At m1AnIn, the maximum value is denoted as fmax1Minimum value is denoted as fmin1(ii) a The following equation is thus obtained:
solving the equation to obtain a1And b1A value of (d);
a2and b2Are respectively the 2 nd market elementAnd the 1 st and 2 nd influence factors of (c), and a is set2>0,b2>0;
At m1AnIn, the maximum value is denoted as fmax2Minimum, isThe value is denoted as fmin2(ii) a The following equation is thus obtained:
solving the equation to obtain a2And b2A value of (d);
similarly, the element V of the investigation result is obtained separatelyi1 influence factor a3And 2 nd influence factor b3A value of (d);
similarly, the academic index factors E are respectively obtainedi1 influence factor a4And 2 nd influence factor b4A value of (d);
for m1Individual added item increment QiAverage value of (2)Obtaining the value f (Q) of the final added value Q of the added items by adopting a statistical distribution algorithm;
and 7, obtaining a final compensation recommendation model as follows:
wherein:
RVfor the grade scoring of the element of the investigation result, the value range is as follows: rV∈(0,1,2,3,4,5,6,7,8,9,10)
REThe grade scores of the academic calendar index elements are in the following value range: rE∈(0,1,2,3,4,5,6,7,8,9,10)
Step 8, adopting the determination of step 2m2And (4) testing the final compensation recommendation model obtained in the step (7) by using the individual candidate sample, and for m2Calculating the amplitude compensation f (Y) of each candidate sample in the candidate samples through a compensation recommendation model1) And characteristic compensation f (Y)2) And by comparing the amplitude compensation f (Y)1) Deviation from its post-entry compensation factor S, and characteristic compensation f (Y)2) And the deviation of the compensation element S after the compensation is performed, and whether the deviation meets the set range, so that a compensation recommendation model is tested;
and 9, after the test in the step 8 is successful, obtaining a salary recommendation result of the candidate j by adopting the following method:
step 9.1, reading compensation basic information of the candidate j; the method comprises the following parameters:
compensation element C before job entryjThe value of (f), (C);
company customization element MjThe value of (f), (M);
the number N of candidate addicts;
step 9.2, determining the grade score R of the element of the investigation result according to the basic information of the candidate jVGrade score R of index elements of the academic calendarE1 rating score R of market elementsIaGrade score R of market item 2Ib(ii) a Reading the fixed value f (Q) of the added value Q of the added item, and obtaining the amplitude compensation f (Y) of the candidate j according to the compensation recommendation model in the step 7j1) And characteristic compensation f (Y)j2);
Step 9.3, according to the following formula, obtaining the compensation recommended range of the candidate j: [ min (f (Y) ]j1),f(Yj1)),max(f(Yj1),f(Yj1)]。
The salary recommendation method based on machine learning provided by the invention has the following advantages:
according to the salary recommendation method based on machine learning, provided by the invention, based on a salary model obtained by machine learning, the interference of human factors on salary is avoided; the compensation method is beneficial to different departments to reach the consistency of compensation of candidates; embodying the personal characteristics of the candidate; the operation is simple, and the full automation can be realized. Thereby improving the effectiveness and efficiency of the recruitment process.
Drawings
Fig. 1 is a flowchart illustrating a compensation recommendation method based on machine learning according to the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects solved by the present invention more clearly apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a machine learning-based compensation recommendation method, which comprises the following steps of referring to fig. 1:
step 1, collecting and establishing a candidate person sample library; the candidate sample library comprises a plurality of candidate samples, and any ith candidate sample is represented as: fi={(f(Ci),Ni,f(Si)};
Wherein:
f(Ci) Represents the pre-entry compensation element C of the ith candidateiValue f (C) ofi);
NiThe number of bonus items representing the ith candidate;
f(Si) Represents the post-entry compensation element S of the ith candidateiValue f (S) ofi);
Step 2, dividing candidate samples in a candidate sample library into a training set and a testing set; wherein the training set has m1A sample of individual candidates; test set has m2A sample of individual candidates;
step 3, for m1The individual candidate sample performs the following compensation model training process:
step 3.1, changing i to 1;
setting initial values, including: setting the 1 st market elementValue ofMarket factor 2Value ofIndex element E of academic calendariValue f (E) ofi) Adding the value added by the item QiValue of (A) f (Q)i) And a factor V of the result of investigationiValue f (V) ofi);
Step 3.2, for the ith candidate sample, reading the pre-entry compensation element C of the ith candidateiValue f (C) ofi) And the compensation element S after workiValue f (S) ofi) Number of bonus items NiAnd company subscription element MiValue f (M)i) (ii) a Wherein, company subscribes element MiValue f (M)i) Is a known fixed value;
step 3.3, adopting an adjusting algorithm to examine the result element ViValue f (V) ofi) Adjusting to obtain an element V of the investigation resultiAnd finally obtaining a survey result element ViAverage value of (2)Element V of the investigation resultiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.3.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.3.2, determine the amplitude compensation f (Y)i1) And characteristic compensation f (Y)i2) Whether the following relation is satisfied:
f(Si)-δ≤f(Yi1)≤f(Si)+δ
f(Si)-δ≤f(Yi2)≤f(Si)+δ
wherein: delta is the market impact coefficient, which is a known set value;
if yes, executing step 3.3.3; if not, executing step 3.3.4;
step 3.3.3, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marked as the significant point, then, changing f (V) by a set adjustment step size + -f (delta V)i) To obtain a new f (V)i) The step 3.3.1 is repeated, and the process is circulated continuously until the set times are reached, and the step 3.3.5 is executed;
step 3.3.4, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marking as an invalid point; then, f (V) is changed by a set adjustment step + -f (Δ V)i) To obtain a new f (V)i) The step 3.3.1 is repeated, and the process is circulated continuously until the set times are reached, and the step 3.3.5 is executed;
step 3.3.5, the distribution of the effective points and the invalid points is counted to obtain the range of the effective point concentration, so that the element V of the investigation result is obtainediThe adjustment range of (a);
Step 3.4, adopting an adjusting algorithm to the 1 st market elementValue ofMaking adjustment to obtain market element 1And finally obtaining the 1 st market elementAverage value of (2)Market 1 elementAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.4.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.4.2, compensate f (Y) according to the amplitude of fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied and further the 1 st market element is satisfiedValue ofAdjusting to obtain the 1 st market element meeting the relation of the step 3.3.2And finally obtaining the 1 st market elementAverage value of (2)
Step 3.5, adopting an adjusting algorithm to the No. 2 market elementValue ofMaking adjustment to obtain the 2 nd market elementAnd finally obtaining the 2 nd market elementAverage value of (2)Market factor 2Average value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.5.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.5.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied or not, and further the 2 nd market elementValue ofAdjusting to obtain the 2 nd market element meeting the relation of the step 3.3.2And finally obtaining the 2 nd market elementAverage value of (2)
Step 3.6, adopting an adjusting algorithm to learn the index element E of the calendariValue f (E) ofi) Regulating to obtain the academic calendar index element EiAnd finally obtaining the academic index element EiAverage value of (2)Will learn the calendar index element EiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.6.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.6.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relation formula of the step 3.3.2 is satisfied or not, and further the academic index element EiValue f (E) ofi) Adjusting to obtain the academic calendar index element E meeting the relational expression of the step 3.3.2iAnd finally obtaining the academic index element EiAverage value of (2)
Step 3.7, adding value Q to the added items by adopting an adjusting algorithmiValue of (A) f (Q)i) Adjusting to obtain the added value QiAnd finally obtaining the added value Q of the added itemiAverage value of (2)Adding the added items to the value QiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.7.1, calculate the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.7.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether step 3.3.2 off is satisfiedSystem, further adding value Q to the added itemiValue of (A) f (Q)i) Adjusting to obtain the added value Q of the added item satisfying the relational expression of the step 3.3.2iAnd finally obtaining the added value Q of the added itemiAverage value of (2)
Step 3.8, so far, the following parameters of the ith candidate are obtained: element V of investigation resultiAverage value of (2)Market 1 elementAverage value of (2)Market factor 2Average value of (2)Index element E of academic calendariAverage value of (2)Sum and add value added amount QiAverage value of (2)
Will be provided withIs assigned to f (V)i),Is assigned toIs assigned toIs assigned to f (E)i),Is assigned to f (Q)i);
Let i equal i + 1; if i is greater than m1Then m is1After the personal candidate samples are trained on the compensation model, executing step 4; otherwise, returning to the step 3.2 for circular execution;
a1and b1Are respectively the 1 st market elementAnd the 1 st and 2 nd influence factors of (c), and a is set1>0,b1>0;
At m1AnIn, the maximum value is denoted as fmax1Minimum value is denoted as fmin1(ii) a The following equation is thus obtained:
solving the equation to obtain a1And b1A value of (d);
a2and b2Are respectively the 2 nd market elementAnd the 1 st and 2 nd influence factors of (c), and a is set2>0,b2>0;
At m1AnIn, the maximum value is denoted as fmax2Minimum value is denoted as fmin2(ii) a The following equation is thus obtained:
solving the equation to obtain a2And b2A value of (d);
similarly, the element V of the investigation result is obtained separatelyi1 influence factor a3And 2 nd influence factor b3A value of (d);
similarly, the academic index factors E are respectively obtainedi1 influence factor a4And 2 nd influence factor b4A value of (d);
for m1Individual added item increment QiAverage value of (2)Obtaining the value f (Q) of the final added value Q of the added items by adopting a statistical distribution algorithm;
and 7, obtaining a final compensation recommendation model as follows:
wherein:
RVfor the grade scoring of the element of the investigation result, the value range is as follows: rV∈(0,1,2,3,4,5,6,7,8,9,10)
RETo grade the academic degree index elements,the value range is as follows: rE∈(0,1,2,3,4,5,6,7,8,9,10)
Step 8, adopting m determined in step 22And (4) testing the final compensation recommendation model obtained in the step (7) by using the individual candidate sample, and for m2Calculating the amplitude compensation f (Y) of each candidate sample in the candidate samples through a compensation recommendation model1) And characteristic compensation f (Y)2) And by comparing the amplitude compensation f (Y)1) Deviation from its post-entry compensation factor S, and characteristic compensation f (Y)2) And the deviation of the compensation element S after the compensation is performed, and whether the deviation meets the set range, so that a compensation recommendation model is tested;
and 9, after the test in the step 8 is successful, obtaining a salary recommendation result of the candidate j by adopting the following method:
step 9.1, reading compensation basic information of the candidate j; the method comprises the following parameters:
compensation element C before job entryjThe value of (f), (C);
company customization element MjThe value of (f), (M);
the number N of candidate addicts;
step 9.2, determining the grade score R of the element of the investigation result according to the basic information of the candidate jVGrade score R of index elements of the academic calendarE1 rating of market elementsRating scores for market items 2Reading the fixed value f (Q) of the added value Q of the added item, and obtaining the amplitude compensation f (Y) of the candidate j according to the compensation recommendation model in the step 7j1) And characteristic compensation f (Y)j2);
Step 9.3, according to the following formula, obtaining the compensation recommended range of the candidate j: [ min (f (Y) ]j1),f(Yj1)),max(f(Yj1),f(Yj1)]。
The salary recommendation method based on machine learning provided by the invention has the following advantages:
(1) establishing a compensation model comprising a plurality of elements, which specifically comprises the following steps: market 1 elementMarket factor 2Index element E of academic calendariAnd element V of investigation resultiThe number of added items NiAnd company customized element Mi(ii) a Wherein, company subscribes element MiIs a fixed value; the value added amount Q of the added items is obtained by training the compensation modeliThe value of (b) is a fixed value, which is applicable to all candidates.
Therefore, when the user uses the trained salary model, the value added by the added items Q only needs to be obtained according to the basic information of the candidateiThe salary before the job, the number of candidate added items and the value of the company customized element, and then the grade scores of each element of the candidate are obtained, so that the salary recommendation range can be quickly obtained. The use is simple and convenient. The finally obtained salary recommendation range is more consistent with the basic information of the candidate through inspection, and has higher credibility.
(2) When training the compensation model, for the 1 st candidate, firstly, the elements V of the investigation result are obtained in sequenceiAverage value of (2)Market 1 elementAverage value of (2)Market factor 2Average value of (2)Index element E of academic calendariAverage value of (2)Sum and add value added amount QiAverage value of (2)In addition, in the case of training the five elements, since the average value of the previous element is used as the known value in the case of training the next element, the accuracy of the model training is enhanced in a progressive relationship.
And starting from the second candidate, each candidate adopts the training result of the previous candidate as the initial known value, so that the training result of the previous candidate is the known value when the next candidate trains, and the accuracy of the model training is further enhanced.
(3) Through the automatic training process of the compensation model, the determined values of the factor influencing each element can be trained. Thereby obtaining an accurate compensation model with universal significance. The salary model carries out salary recommendation on a plurality of interviewees, and the accuracy and the reliability of the use result of the salary model are verified through a large amount of data verification and evaluation of a plurality of experts.
(4) For compensation information, further mining can be performed:
1) two compensation calculation results obtained by final calculation, namely amplitude compensation f (Y)1) And characteristic compensation f (Y)2) The closer the numbers, the more normal the candidate is developing;
2) when amplitude salary f (Y)1) Is obviously greater than the characteristic compensation f (Y)2) When the time is short, the development of the candidate is beyond the general situation;
3) characteristic compensation f (Y)2) Is obviously greater than the amplitude compensation f (Y)1) Time, indicates that the candidate is developing less than normal.
In the latter two cases, the compensation model may be further trained by increasing the number of training samples.
The compensation model obtained by the final training of the invention adopts excel expression, and when in use, the compensation model can be directly used for filling in a form, thus being simple and convenient
Taking a compensation model of a king of a certain candidate as an example, through calculation, the amplitude compensation calculation is 12279 yuan, and the characteristic compensation calculation is 11633 yuan, so the reasonable compensation range is [11633,12279], the optimal compensation is 11956 yuan, so the recommended compensation is 12000 yuan, and the amplitude is 19%.
The invention provides a set of analysis methods for training of complete compensation influence factors, which can reflect compensation characteristics of interview candidates and the contribution degree of each characteristic in compensation suggestions, and finally generate a reasonable compensation suggestion range.
The invention provides a machine learning-based salary recommendation method, which has the following advantages:
(1) interference of human factors on compensation is avoided based on a compensation model obtained by machine learning; the compensation method is beneficial to different departments to reach the consistency of compensation of candidates; embodying the personal characteristics of the candidate; the operation is simple, and the full automation can be realized. Thereby improving the effectiveness and efficiency of the recruitment process.
(2) The model provides a machine learning-based salary factor model and a factor training method, and calculates salary range and advice salary according to two different dimensions, and is used for guiding salary amount evaluation in the recruitment process.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and improvements can be made without departing from the principle of the present invention, and such modifications and improvements should also be considered within the scope of the present invention.
Claims (1)
1. A compensation recommendation method based on machine learning is characterized by comprising the following steps:
step 1, collecting and establishing a candidate person sample library; the candidate sample library comprises a plurality of candidate samples, and any ith candidate sample is represented as: fi={(f(Ci),Ni,f(Si)};
Wherein:
f(Ci) Represents the pre-entry compensation element C of the ith candidateiValue f (C) ofi);
NiThe number of bonus items representing the ith candidate;
f(Si) Represents the post-entry compensation element S of the ith candidateiValue f (S) ofi);
Step 2, dividing candidate samples in a candidate sample library into a training set and a testing set; wherein the training set has m1A sample of individual candidates; test set has m2A sample of individual candidates;
step 3, for m1The individual candidate sample performs the following compensation model training process:
step 3.1, changing i to 1;
setting initial values, including: setting the 1 st market elementValue ofMarket factor 2Value ofIndex element E of academic calendariValue f (E) ofi) Adding the value added by the item QiValue of (A) f (Q)i) And a factor V of the result of investigationiValue f (V) ofi);
Step 3.2, for the ith candidate sample, reading the pre-entry compensation element C of the ith candidateiValue f (C) ofi) And the compensation element S after workiValue f (S) ofi) Number of bonus items NiAnd company subscription element MiValue f (M)i) (ii) a Wherein, company subscribes element MiValue f (M)i) Is a known fixed value;
step 3.3, adopting an adjusting algorithm to examine the result element ViValue f (V) ofi) Adjusting to obtain an element V of the investigation resultiAnd finally obtaining a survey result element ViAverage value of (2)Element V of the investigation resultiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.3.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.3.2, determine the amplitude compensation f (Y)i1) And characteristic compensation f (Y)i2) Whether the following relation is satisfied:
f(Si)-δ≤f(Yi1)≤f(Si)+δ
f(Si)-δ≤f(Yi2)≤f(Si)+δ
wherein: delta is the market impact coefficient, which is a known set value;
if yes, executing step 3.3.3; if not, executing step 3.3.4;
step 3.3.3, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marked as the significant point, then, changing f (V) by a set adjustment step size + -f (delta V)i) To obtain a new f (V)i) Then returning to the step 3.3.1, and continuously doing soCirculating until the set times is reached, and executing the step 3.3.5;
step 3.3.4, the element V of the investigation result of the current i-th candidate is usediValue f (V) ofi) Marking as an invalid point; then, f (V) is changed by a set adjustment step + -f (Δ V)i) To obtain a new f (V)i) The step 3.3.1 is repeated, and the process is circulated continuously until the set times are reached, and the step 3.3.5 is executed;
step 3.3.5, the distribution of the effective points and the invalid points is counted to obtain the range of the effective point concentration, so that the element V of the investigation result is obtainediThe adjustment range of (a);
Step 3.4, adopting an adjusting algorithm to the 1 st market elementValue ofMaking adjustment to obtain market element 1And finally obtaining the 1 st market elementAverage value of (2)Market 1 elementAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.4.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.4.2, compensate f (Y) according to the amplitude of fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied and further the 1 st market element is satisfiedValue ofAdjusting to obtain the 1 st market element meeting the relation of the step 3.3.2And finally obtaining the 1 st market elementAverage value of (2)
Step 3.5, adopting an adjusting algorithm to the No. 2 market elementValue ofMaking adjustment to obtain the 2 nd market elementAnd finally obtaining the 2 nd market elementAverage value of (2)Market factor 2Average value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.5.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.5.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relational expression of the step 3.3.2 is satisfied or not, and further the 2 nd market elementValue ofAdjusting to obtain the satisfied stepStep 3.3.2 market elements of the relation 2And finally obtaining the 2 nd market elementAverage value of (2)
Step 3.6, adopting an adjusting algorithm to learn the index element E of the calendariValue f (E) ofi) Regulating to obtain the academic calendar index element EiAnd finally obtaining the academic index element EiAverage value of (2)Will learn the calendar index element EiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.6.1, respectively calculating and obtaining the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.6.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relation formula of the step 3.3.2 is satisfied or not, and further the academic index element EiValue f (E) ofi) Adjusting to obtain the academic calendar index elements meeting the relational expression of the step 3.3.2EiAnd finally obtaining the academic index element EiAverage value of (2)
Step 3.7, adding value Q to the added items by adopting an adjusting algorithmiValue of (A) f (Q)i) Adjusting to obtain the added value QiAnd finally obtaining the added value Q of the added itemiAverage value of (2)Adding the added items to the value QiAverage value of (2)Recording the data in a result table;
wherein, the adjusting algorithm is as follows:
step 3.7.1, calculate the amplitude compensation f (Y) according to the following formulai1) And characteristic compensation f (Y)i2):
Step 3.7.2, compensate f (Y) according to the amplitude fluctuationi1) And characteristic compensation f (Y)i2) Whether the relation formula of the step 3.3.2 is satisfied or not, and further adding the value Q to the added itemsiValue of (A) f (Q)i) Adjusting to obtain the added value Q of the added item satisfying the relational expression of the step 3.3.2iAnd finally obtaining the added value Q of the added itemiAverage value of (2)
Step 3.8, thus obtaining the ithCandidates the following parameters: element V of investigation resultiAverage value of (2)Market 1 elementAverage value of (2)Market factor 2Average value of (2)Index element E of academic calendariAverage value of (2)Sum and add value added amount QiAverage value of (2)
Will be provided withIs assigned to f (V)i),Is assigned toIs assigned toIs assigned to f (E)i),Is assigned to f (Q)i);
Let i equal i + 1; if i is greater than m1Then m is1After the personal candidate samples are trained on the compensation model, executing step 4; otherwise, returning to the step 3.2 for circular execution;
a1and b1Are respectively the 1 st market elementAnd the 1 st and 2 nd influence factors of (c), and a is set1>0,b1>0;
At m1AnIn, the maximum value is denoted as fmax1Minimum value is denoted as fmin1(ii) a The following equation is thus obtained:
solving the equation to obtain a1And b1A value of (d);
a2and b2Are respectively the 2 nd market elementAnd the 1 st and 2 nd influence factors of (c), and a is set2>0,b2>0;
At m1AnIn, the maximum value is denoted as fmax2Minimum value is denoted as fmin2(ii) a The following equation is thus obtained:
solving the equation to obtain a2And b2A value of (d);
similarly, the element V of the investigation result is obtained separatelyi1 influence factor a3And 2 nd influence factor b3A value of (d);
similarly, the academic index factors E are respectively obtainedi1 influence factor a4And 2 nd influence factor b4A value of (d);
for m1Individual added item increment QiAverage value of (2)Obtaining the value f (Q) of the final added value Q of the added items by adopting a statistical distribution algorithm;
and 7, obtaining a final compensation recommendation model as follows:
wherein:
RVfor the grade scoring of the element of the investigation result, the value range is as follows: rV∈(0,1,2,3,4,5,6,7,8,9,10)
RERating for academic index elementsThe value range is: rE∈(0,1,2,3,4,5,6,7,8,9,10)
Step 8, adopting m determined in step 22And (4) testing the final compensation recommendation model obtained in the step (7) by using the individual candidate sample, and for m2Calculating the amplitude compensation f (Y) of each candidate sample in the candidate samples through a compensation recommendation model1) And characteristic compensation f (Y)2) And by comparing the amplitude compensation f (Y)1) Deviation from its post-entry compensation factor S, and characteristic compensation f (Y)2) And the deviation of the compensation element S after the compensation is performed, and whether the deviation meets the set range, so that a compensation recommendation model is tested;
and 9, after the test in the step 8 is successful, obtaining a salary recommendation result of the candidate j by adopting the following method:
step 9.1, reading compensation basic information of the candidate j; the method comprises the following parameters:
compensation element C before job entryjThe value of (f), (C);
company customization element MjThe value of (f), (M);
the number N of candidate addicts;
step 9.2, determining the grade score R of the element of the investigation result according to the basic information of the candidate jVGrade score R of index elements of the academic calendarE1 rating score R of market elementsIaGrade score R of market item 2Ib(ii) a Reading the fixed value f (Q) of the added value Q of the added item, and obtaining the amplitude compensation f (Y) of the candidate j according to the compensation recommendation model in the step 7j1) And characteristic compensation f (Y)j2);
Step 9.3, according to the following formula, obtaining the compensation recommended range of the candidate j: [ min (f (Y) ]j1),f(Yj1)),max(f(Yj1),f(Yj1)]。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011265168.7A CN112381510A (en) | 2020-11-12 | 2020-11-12 | Salary recommendation method based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011265168.7A CN112381510A (en) | 2020-11-12 | 2020-11-12 | Salary recommendation method based on machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112381510A true CN112381510A (en) | 2021-02-19 |
Family
ID=74583602
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011265168.7A Pending CN112381510A (en) | 2020-11-12 | 2020-11-12 | Salary recommendation method based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112381510A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112561500A (en) * | 2021-02-25 | 2021-03-26 | 深圳平安智汇企业信息管理有限公司 | Salary data generation method, device, equipment and medium based on user data |
-
2020
- 2020-11-12 CN CN202011265168.7A patent/CN112381510A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112561500A (en) * | 2021-02-25 | 2021-03-26 | 深圳平安智汇企业信息管理有限公司 | Salary data generation method, device, equipment and medium based on user data |
CN112561500B (en) * | 2021-02-25 | 2021-05-25 | 深圳平安智汇企业信息管理有限公司 | Salary data generation method, device, equipment and medium based on user data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Mattick et al. | Approaches to learning and studying in medical students: validation of a revised inventory and its relation to student characteristics and performance | |
KR100919912B1 (en) | Systems and methods for semantic knowledge assessment, instruction, and acquisition | |
US20220261655A1 (en) | Real-time prediction method for engine emission | |
CN112381510A (en) | Salary recommendation method based on machine learning | |
CN114049240A (en) | Method for intelligently recommending questions and learning paths based on student online learning data | |
CN117574201A (en) | Model training method, device, equipment and storage medium based on multi-industry model | |
CN116579795A (en) | Municipal construction engineering cost evaluation system based on big data | |
CN110851784A (en) | Early warning method for field operation of electric energy meter | |
CN116596395B (en) | Operation quality control platform for engineering project evaluation unit guidance and detection | |
CN115375210B (en) | Occupational skill level identification management method and system | |
CN117313947A (en) | Learning early warning method based on random forest and tuna group optimization algorithm | |
CN111639194A (en) | Knowledge graph query method and system based on sentence vectors | |
Imbrie et al. | Comparison of four methodologies for modeling student retention in engineering | |
Şengül Avşar | Aberrant individuals’ effects on fit indices both of confirmatory factor analysis and polytomous IRT models | |
CN115409257A (en) | Score distribution prediction method and system based on condition density estimation model | |
WO2022011827A1 (en) | Method for optimizing loss function by means of error cause reinforcement | |
CN114048675B (en) | Hysteresis motor component coupling method | |
CN112579866B (en) | Method for analyzing event development trend based on time heat index | |
CN115221863B (en) | Text abstract evaluation method, device and storage medium | |
Balona et al. | The actuary and IBNR Techniques | |
CN111707795B (en) | Method and system for monitoring Biochemical Oxygen Demand (BOD) of surface water | |
CN114386196B (en) | Method for evaluating mechanical property prediction accuracy of plate strip | |
Sun et al. | Research on Predictive Control of Students' Achievements in Physical Education Class and Physical Education Class Universities Based on Mathematical Model Analysis of Multiple Linear Regression Equations | |
Zhen et al. | Factors Affecting Millennial College Students’ Job Satisfaction in Jiangsu Province in China | |
CN118229151A (en) | Occupational bias evaluation method for index type summarization analysis |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210219 |
|
RJ01 | Rejection of invention patent application after publication |