CN106295969A - Power customer is worth a kind of weighting K means method hived off - Google Patents
Power customer is worth a kind of weighting K means method hived off Download PDFInfo
- Publication number
- CN106295969A CN106295969A CN201610623649.8A CN201610623649A CN106295969A CN 106295969 A CN106295969 A CN 106295969A CN 201610623649 A CN201610623649 A CN 201610623649A CN 106295969 A CN106295969 A CN 106295969A
- Authority
- CN
- China
- Prior art keywords
- power customer
- customers
- data
- weighting
- client
- 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 22
- 238000003064 k means clustering Methods 0.000 claims abstract description 19
- 230000000694 effects Effects 0.000 claims abstract description 18
- 208000024780 Urticaria Diseases 0.000 claims abstract description 16
- 238000005259 measurement Methods 0.000 claims description 2
- 230000005611 electricity Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000007418 data mining Methods 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 230000036528 appetite Effects 0.000 description 2
- 235000019789 appetite Nutrition 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011985 exploratory data analysis Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Primary Health Care (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of a kind of weighting K means method being worth for power customer and hiving off.Present invention uses a kind of weighting K means clustering algorithm being suitable for power customer characteristic variable data characteristics, first algorithm determines in the power customer group of weighting the standard deviation sum of data for clustering criteria function, and weight is the ratio that client's number accounts for total client's number in this power customer group;Then according to this criterion function when the similarity calculated between power customer object and customers' central point, premised on Euclidean distance, imposing data standard difference in customers is the weight of reference factor, it is achieved the power customer object of density unevenness is worth more accurately and hives off.Weighting K means clustering algorithm applies the cluster result that hives off on power customer value is hived off to show, the present invention is to be suitable for actual operation data, and has reached to improve the effect of cluster compactedness.More the cluster result that hives off of high-quality is it is also ensured that decision-making efficient implementation, finally brings higher income for power supply enterprise.
Description
Technical field
The present invention relates to power customer and be worth a kind of clustering method hived off, be specifically related to power customer and be worth hived off
Plant weighting K-means method.
Background technology
21 century is the epoch of an information, and information all serves a vital work for the impact of all trades and professions
With.In the face of current power supply enterprise, every day is all producing and the huge management of enterprise operation data of renewal, then how will go profit
Use these data, from numerous in disorder data, excavate potential customer value, and then help electric power enterprise to improve marketing certainly
Plan, cut operating costs, improve enterprise income, be that each power supply enterprise is in the direction made great efforts.Data mining technology is as one
Planting and can find in mass data that the data processing means of potential information is just shown one's talent at this, this technology has become place
The important means of the mass historical data accumulated in the process of construction of reason electric power trade information, the application of data mining technology
Also a development space the most wide will be provided for power supply enterprise.
Tradition K-means clustering algorithm is a kind of data mining technology means that current customer grouping is conventional.Tradition K-
Means clustering algorithm is simple, and convergence rate the most quickly, is therefore typically used to hiving off of client.Its way is to give birth to the most at random
Become K initial cluster center, then remaining data sample be included into group belonging to K cluster centre, recalculate cluster centre,
If center changes, the most again cluster, until cluster centre does not changes, terminate algorithm.
For the feature that power customer characteristic variable data distribution density difference is big, if directly using tradition K-means
Clustering algorithm, it is clear that high density groupuscule can be caused to carve up the phenomenon of low-density jumpbogroup.
Summary of the invention
The technical problem to be solved in the present invention is: proposes a kind of power customer and is worth a kind of weighting K-means side hived off
Method, it is suitable for actual operation data, and has reached to improve the effect of cluster compactedness, it is ensured that decision-making efficient implementation.
A kind of weighting K-means method hived off it is worth for power customer, it is characterised in that: become from power customer feature
The feature that amount data distribution density difference is big is set out, and uses the K-means clustering algorithm of weighting that power customer is realized have titer
Value is hived off, and specifically includes following steps:
Step 1: in first algorithm determines the power customer group to weight, the standard deviation sum of data is for clustering criteria letter
Number, weight is the ratio that client's number accounts for total client's number in this power customer group;
Step 2: then according to this criterion function in the similarity calculated between power customer object and customers' central point
Time, premised on Euclidean distance, imposing data standard difference in customers is the weight of reference factor, it is achieved the electric power of density unevenness
Target client is worth more accurately and hives off.
With weighting power customer group's internal standard difference sum for clustering criteria function, be Clustering Effect that power customer is hived off
Measurement;First calculate the standard deviation of data in each power customer group of grouping result, and be aided with this to corresponding standard deviation
In group, client's number accounts for the ratio of total client's number as weight, and the effect of weight is to increase the more customers' standard deviation of client's number
Contribution degree, the summation of the standard deviation finally each weighted is hived off measure of effectiveness criterion as final power customer.
According to the clustering criteria function of weighting, in the similarity calculated between power customer object and customers' central point
Time, premised on Euclidean distance, imposing data standard difference in customers is the weight of reference factor;Calculate target client and client
During similarity between group center's point, first calculate the Euclidean distance between target client and customers' central point, basis at this
On, imposing the inverse of data standard difference evolution in the group originally of customers centers is weight, using this weighted euclidean distance as visitor
The tolerance of the similarity between family object and customers' central point, the effect of weight is to increase density low jumpbogroup data Euclidean distance
Contribution degree, the mistake point situation of border client between high density groupuscule and low-density jumpbogroup can be reduced so that scattered client
It is attributed to sparse customers.
There is advantages that present invention uses one is suitable for power customer characteristic variable data characteristics
Weighting K-means clustering algorithm, first algorithm determines the standard deviation sum of data in the power customer group of weighting is poly-
Class criterion function, weight is the ratio that client's number accounts for total client's number in this power customer group;Then according to this criterion function
When the similarity calculated between power customer object and customers' central point, premised on Euclidean distance, impose in customers
Data standard difference is the weight of reference factor, it is achieved the power customer object of density unevenness is worth more accurately and hives off.Weighting K-
Means clustering algorithm applies the cluster result that hives off on power customer value is hived off to show, the present invention is to be suitable for reality operation
Data, and reached to improve the effect of cluster compactedness.More the cluster result that hives off of high-quality is it is also ensured that decision-making is high
Effect is implemented, and finally brings higher income for power supply enterprise.
Accompanying drawing explanation
Fig. 1 is the flow chart of the present invention;
Fig. 2 is the flow chart of step 1 data prediction of the present invention;
Fig. 3 is Very Important Person lifting degree figure;
Fig. 4 is big customer's degree of lifting figure;
Fig. 5 attaches most importance to and pays close attention to client's degree of lifting figure;
Fig. 6 is residential customers lifting degree figure;
Fig. 7 is other client's degree of lifting figures.
Detailed description of the invention
The present invention is described in further detail with embodiment below in conjunction with the accompanying drawings.
Power customer is worth the weighting K-means clustering method process hived off, and specifically comprises the following steps that
Step 1: data prediction.First raw power client's marketing data is carried out exploratory analysis, on this basis,
Reject the variable unrelated with analyzing target or extract variable needed for (structure) model, at these data selected
Reason.Converted by the cleaning of power customer marketing data, data integration and data, raw power client's marketing data is processed into mould
Input feature vector data set required for type.
Step 2: the power customer characteristic variable data set having handled step 1 well clusters for the first time and hives off.First by
Random method chooses K initial cluster center, uses the shortest principle of Euclidean distance, by remaining N-K power customer object
Being included into the group adhering to K cluster centre separately, K the power customer completing to cluster for the first time is worth group.According to the method weight calculating average
The new cluster centre calculating K power customer value group, if central point changes, carries out the iteration of step 3, otherwise eventually
Only algorithm, exports cluster result.
Step 3: if step 2 does not terminate, then use weighted euclidean distance regrouping cluster, and this process is innovation
Iterative process.First calculate power customer belonging to cluster centre according to a front grouping result and be worth the standard deviation of group's data, with
The inverse of this standard deviation evolution is the weight of corresponding power customer cluster centre, calculates power customer object difference beyond a good appetite suddenly appearing in a serious disease heart
To the weighted euclidean distance of K cluster centre, use the principle that weighted euclidean distance is the shortest, the power customer object that will do not hive off
It is included into and adheres to K cluster centre place group separately, complete iteration and hive off cluster.Then recalculate with the method for calculating average and obtain newly
Cluster centre, if center changes, repeat the iterative process of this step 3, otherwise termination algorithm, export result.
Step 4: model evaluation.The degree of accuracy of model is weighed from the compactedness of model and lifting degree.The compactedness of model
The standard deviation using power customer grouping result is weighed, and the lifting degree of model uses tradition K-means cluster and weighting K-
The slip of means cluster result standard deviation is weighed.
Described step 1: be described as follows:
2 months (in August, 2015 in JIUYUE ,-2015) power customer operation datas of extraction Kweiyang power supply enterprise are as original
Data set.Analyze through Data Mining, it is determined that modeling raw data set, a total of 6078801 records of its quantity, from this
To carrying out data cleansing (as deleted the paid electricity charge record data less than 0), data conversion (as to industry code weight in data set
Newly encoded), structure's variable (such as monthly power consumption=total electricity consumption/number of times should be paid), variable standardization.The modeling finally determined refers to
Mark variable include monthly power consumption, average electricity price, on schedule rate of redemption, averagely pay the fees duration, credit score, electricity consumption class code,
Trade classification code, contract capacity, index eliminates dimension standardisation process and uses extreme difference standardization:Wherein, X ' represents the power customer characteristic variable data after extreme difference standardization, and X represents original
Power customer characteristic variable data, XmaxRepresent power customer characteristic variable data maximums, contrary XminRepresent power customer special
Levy the minima of variable data.2516721 records are extracted, as Experimental modeling data set after pretreatment.In view of different
The magnitude differences of customer class variable data value is very big, and the difference such as residential customers and the power consumption of big customer and the electricity charge is all very
Big, the value brought to reduce this variate-value difference is hived off error, and power customer is divided into 5 big classes by this experiment, including
Very Important Person, big customer, customer requiring extraordinary attention, residential customers, other clients, carry out customer value respectively and hive off cluster.
Described step 2 is described as follows:
The power customer characteristic variable data set having handled step 1 well clusters for the first time and hives off.First by random
Method chooses K power customer object as initial cluster center, uses the shortest principle of Euclidean distance, by remaining N-K electric
Power target client is included into the group adhering to K power customer cluster centre separately, and K the power customer completing to cluster for the first time is worth group.Root
Recalculate K power customer according to the method calculating average to be worth the cluster centre of group (this center can be objective with certain electric power of right and wrong
The data at family, belong to the concept of average), if the central point of customer value group changes, carry out the iteration of step 3, otherwise
Termination algorithm, output power customer value hives off cluster result.
The computing formula of Euclidean distance is:
Wherein, d represents Euclidean distance, and q represents the number of power customer value characteristic variable, and p represents pth power customer
Value characteristic variable, x1、x2Represent two power customer objects.
Described step 3 is described as follows:
If step 2 does not terminate, then using weighted euclidean distance regrouping cluster, this process is the iteration of innovation
Process.First it is worth grouping result according to a front power customer and calculates the mark of power customer value group's data belonging to cluster centre
Accurate poor, with the inverse of this standard deviation evolution for the weight of corresponding power customer cluster centre, calculate power customer beyond a good appetite suddenly appearing in a serious disease heart
Object arrives the weighted euclidean distance of K cluster centre respectively, uses the principle that weighted euclidean distance is the shortest, the electric power that will do not hive off
Target client is included into and adheres to K cluster centre place group separately, completes iteration and hives off cluster.Then by calculate average method again in terms of
Calculate the cluster centre obtaining new power customer value grouping result, if center changes, repeat the iteration mistake of this step 3
Journey, otherwise termination algorithm, output power customer value hives off cluster result.
Weighted euclidean distance computing formula: Wk·dist(Ck·x)
Wherein dist () represents the Euclidean distance of data between calculating power customer object;CkRepresent kth power customer
It is worth the cluster centre of group;X represents any power customer object except cluster centre;WkRepresent kth power customer and be worth group
Weight corresponding to cluster centre, i.e. the inverse of the standard deviation evolution of customers' data that this cluster centre is original
The computing formula of standard deviation:
Wherein, n represents this power customer and is worth total client's number of group, xiRepresent that this power customer is worth a client of group
Data object, μ represents that this power customer is worth the data mean value of group.
Described step 4 is described as follows:
The degree of accuracy of model is weighed from the compactedness of model and lifting degree.The compactedness of model uses power customer to hive off
The standard deviation of result is weighed, and the lifting degree of model uses tradition K-means cluster and weighting K-means cluster result standard deviation
Slip weigh.
The computing formula of standard deviation is given by step 3, and standard deviation the least explanation power customer is worth the compactedness hived off
The strongest, the Clustering Effect that hives off is the best.
The computing formula of lifting degree: r=[(σ '-σ)/σ '] × 100%
Wherein, r represents lifting degree, is also that weighting K-means clustering algorithm compares tradition K-means clustering algorithm standard deviation
Slip, it is poor that σ represents that the weighting power customer that obtains of K-means clustering algorithm is worth the data standard of grouping result, σ ' table
On the premise of showing that traditional K-means clustering algorithm uses same initial cluster center, the power customer obtained is worth and hives off
The data standard of result is poor.Lifting degree is than traditional K-means algorithm cluster effect on the occasion of explanation weighting K-means clustering algorithm
Really compactedness is more preferable, for illustrating during negative value that effect is worse;The value the biggest explanation weighting K-means of lifting degree is than traditional K-means
Clustering Effect is the best, and more novel obvious results fruit is the poorest.
As it can be seen from table 1 the K-means clustering algorithm of weighting makes power customer be worth all of the cluster result that hives off
The meansigma methods of customer value group's standard deviation substantially reduces, and customer value group's standard deviation of 5 customer class averagely reduces
14.50%, the K-means clustering algorithm of this explanation weighting makes power customer be worth each customer value group of cluster of hiving off more
For compact.Especially, the value of residential customers is hived off in cluster result, and the standard deviation of all customers all decreases, slip
Scope is 4.88%-96.00%, hence it is evident that improve the effect being worth cluster of hiving off.Although other 4 customer class all can occur
The standard deviation having customers is deteriorated, but has promoted the standard deviation of other customers significantly more to improve, thus ensures
The compactedness of overall effect of hiving off.The K-means clustering algorithm of weighting is
Cost, to guarantee that the overall value of power customer is hived off the improvement of Clustering Effect.And, the most more open power customer
Object is exactly to compare ambiguous client, it is likely that be exactly that a more open power customer is worth group, or noise
The power customer object of point, exceptional value etc, therefore power customer big for density is worth the power customer that group's periphery is more open
Object is divided into loose power customer and is worth group, it is ensured that power customer overall value is hived off the improvement of Clustering Effect.
Table 1 tradition K-means compares with weighting K-means result each group standard deviation
Certainly, being more than the concrete exemplary applications of the present invention, the present invention also has other embodiment, and all employings are equal to
Replace or the technical scheme of equivalent transformation formation, within all falling within protection domain of the presently claimed invention.
Claims (3)
1. one kind is worth a kind of weighting K-means method hived off for power customer, it is characterised in that: from power customer feature
The feature that variable data distribution density difference is big is set out, and uses the K-means clustering algorithm of weighting to realize power customer effectively
Value is hived off, and specifically includes following steps:
Step 1: in first algorithm determines the power customer group to weight, the standard deviation sum of data is for clustering criteria function, power
The ratio of total client's number is heavily accounted for for client's number in this power customer group;
Step 2: then according to this criterion function when the similarity calculated between power customer object and customers' central point, with
Premised on Euclidean distance, imposing data standard difference in customers is the weight of reference factor, it is achieved the power customer of density unevenness
Object is worth more accurately and hives off.
Feature based the most according to claim 1 selects the LR-Bagging algorithm improved, it is characterised in that: with weighting
Power customer group's internal standard difference sum is clustering criteria function, is the measurement of Clustering Effect of hiving off power customer;First calculate
The standard deviation of data in each power customer group of grouping result, and client's number accounts for total visitor in being aided with this group to corresponding standard deviation
The ratio of amount is as weight, and the effect of weight is to increase the contribution degree of the more customers' standard deviation of client's number, finally will be each
The summation of the standard deviation of individual weighting is hived off measure of effectiveness criterion as final power customer.
Feature based the most according to claim 1 selects the LR-Bagging algorithm improved, it is characterised in that: according to weighting
Clustering criteria function, when the similarity calculated between power customer object and customers' central point, be front with Euclidean distance
Carrying, imposing data standard difference in customers is the weight of reference factor;Calculate the phase between target client and customers central points
When seemingly spending, first calculate the Euclidean distance between target client and customers' central point, on this basis, impose customers center
In group originally, the inverse of data standard difference evolution is weight, using this weighted euclidean distance as target client and customers center
The tolerance of the similarity between point, the effect of weight is to increase the contribution degree of density low jumpbogroup data Euclidean distance, can reduce
Between high density groupuscule and low-density jumpbogroup, the mistake of border client divides situation so that scattered client is attributed to sparse customers.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610623649.8A CN106295969A (en) | 2016-08-02 | 2016-08-02 | Power customer is worth a kind of weighting K means method hived off |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610623649.8A CN106295969A (en) | 2016-08-02 | 2016-08-02 | Power customer is worth a kind of weighting K means method hived off |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106295969A true CN106295969A (en) | 2017-01-04 |
Family
ID=57664171
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610623649.8A Pending CN106295969A (en) | 2016-08-02 | 2016-08-02 | Power customer is worth a kind of weighting K means method hived off |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106295969A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108154312A (en) * | 2018-01-17 | 2018-06-12 | 河南工业大学 | A kind of method for building weight coefficient overall merit wheat preservation quality |
CN109525337A (en) * | 2017-09-20 | 2019-03-26 | 腾讯科技(深圳)有限公司 | WiFi fingerprint acquisition methods, device, storage medium and equipment |
CN110175468A (en) * | 2019-05-05 | 2019-08-27 | 浙江工业大学 | A kind of name desensitization method retaining distribution characteristics |
CN111159258A (en) * | 2019-12-31 | 2020-05-15 | 科技谷(厦门)信息技术有限公司 | Customer clustering implementation method based on cluster analysis |
CN113111924A (en) * | 2021-03-26 | 2021-07-13 | 邦道科技有限公司 | Electric power customer classification method and device |
CN116028838A (en) * | 2023-01-09 | 2023-04-28 | 广东电网有限责任公司 | Clustering algorithm-based energy data processing method and device and terminal equipment |
CN117493979A (en) * | 2023-12-29 | 2024-02-02 | 青岛智简尚达信息科技有限公司 | Customer classification method based on data processing |
-
2016
- 2016-08-02 CN CN201610623649.8A patent/CN106295969A/en active Pending
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109525337A (en) * | 2017-09-20 | 2019-03-26 | 腾讯科技(深圳)有限公司 | WiFi fingerprint acquisition methods, device, storage medium and equipment |
CN108154312A (en) * | 2018-01-17 | 2018-06-12 | 河南工业大学 | A kind of method for building weight coefficient overall merit wheat preservation quality |
CN110175468A (en) * | 2019-05-05 | 2019-08-27 | 浙江工业大学 | A kind of name desensitization method retaining distribution characteristics |
CN111159258A (en) * | 2019-12-31 | 2020-05-15 | 科技谷(厦门)信息技术有限公司 | Customer clustering implementation method based on cluster analysis |
CN113111924A (en) * | 2021-03-26 | 2021-07-13 | 邦道科技有限公司 | Electric power customer classification method and device |
CN116028838A (en) * | 2023-01-09 | 2023-04-28 | 广东电网有限责任公司 | Clustering algorithm-based energy data processing method and device and terminal equipment |
CN116028838B (en) * | 2023-01-09 | 2023-09-19 | 广东电网有限责任公司 | Clustering algorithm-based energy data processing method and device and terminal equipment |
CN117493979A (en) * | 2023-12-29 | 2024-02-02 | 青岛智简尚达信息科技有限公司 | Customer classification method based on data processing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106295969A (en) | Power customer is worth a kind of weighting K means method hived off | |
CN109063945B (en) | Value evaluation system-based 360-degree customer portrait construction method for electricity selling company | |
CN106600455A (en) | Electric charge sensitivity assessment method based on logistic regression | |
CN105160416A (en) | Transformer area reasonable line loss prediction method based on principal component analysis and neural network | |
Farajian et al. | Mining the banking customer behavior using clustering and association rules methods | |
CN106779203B (en) | Plateau mountain area wind power prediction method based on different wind speed sections | |
CN106897402A (en) | The method and user's portrait maker of user's portrait are built based on social security data | |
CN108388974A (en) | Top-tier customer Optimum Identification Method and device based on random forest and decision tree | |
CN109766950A (en) | A kind of industrial user's short-term load forecasting method based on form cluster and LightGBM | |
CN107274066B (en) | LRFMD model-based shared traffic customer value analysis method | |
CN108389069A (en) | Top-tier customer recognition methods based on random forest and logistic regression and device | |
CN110119948A (en) | Based on when variable weight dynamic combined power consumer credit assessment method and system | |
CN108230029A (en) | Client trading behavior analysis method | |
CN103325067A (en) | Service recommendation method and system based on electricity customer segmentation | |
CN108364191A (en) | Top-tier customer Optimum Identification Method and device based on random forest and logistic regression | |
CN108154311A (en) | Top-tier customer recognition methods and device based on random forest and decision tree | |
CN106777005A (en) | User power utilization behavior analysis method based on big data technological improvement clustering algorithm | |
CN108280562A (en) | A kind of method of specification electric power enterprise data resource | |
CN106202388B (en) | A kind of user gradation Automated Partition Method and system | |
CN104992279A (en) | Performance assessment method and device | |
CN115310752A (en) | Energy big data-oriented data asset value evaluation method and system | |
CN104751369A (en) | Improved line loss analysis method for K-means clustering | |
CN107239853B (en) | Intelligent housekeeper system based on cloud computing and working method thereof | |
CN107578130A (en) | A kind of method of the copper forecasting of futures prix of the TSK fuzzy systems of structure and parameter combination learning | |
Luo et al. | China's lagging region development and targeted transportation infrastructure investments |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170104 |