CN110046934A - A kind of idle clothing valuation system and method based on data mining - Google Patents
A kind of idle clothing valuation system and method based on data mining Download PDFInfo
- Publication number
- CN110046934A CN110046934A CN201910283617.1A CN201910283617A CN110046934A CN 110046934 A CN110046934 A CN 110046934A CN 201910283617 A CN201910283617 A CN 201910283617A CN 110046934 A CN110046934 A CN 110046934A
- Authority
- CN
- China
- Prior art keywords
- price
- data
- characteristic
- clothing
- idle
- 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
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0206—Price or cost determination based on market factors
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of idle clothing valuation system and method based on data mining, this method obtains the characteristic group of idle clothing, dimensionization processing is carried out to the non-quantized characteristic for influencing price, obtains each product features to the feature weight of the second-hand price of commodity, obtain the mode of the second-hand calculation of price of idle clothing, the second-hand goods value expectation that both parties meet marketability is given, value cognition of the balance both sides in transaction is poor.And then improve the experience of both parties' two-way trade.
Description
Technical field
The present invention relates to value assessment field, in particular to a kind of idle clothing valuation system and side based on data mining
Method.
Background technique
With the development of internet industry, more and more pass-along deal platforms come into being, and mainstream is second-hand on the market
It the leisured fish of transaction platform and walks around, in not busy fish and walks around upper it can be seen that there is reselling for various idle clothings, however both
Transaction platform, still without implementing mature systems approach, leads to idle clothing in terms of the price expectation and tasting to idle clothing
The case where there are information asymmetries in the market of reselling of object, and then affect the balance of second-hand clothing trade market.It seem second-hand object
Pass-along deal price of the product on not busy fish is independently fixed a price by the seller, is thus easily lead to pass-along deal price and is deviateed reality
Market value, so as to cause the damage in interests is received either in the seller and buyer;It seem to walk around also only in mobile phone
Etc. several moneys sell fast type electronic product on using third party tasting by the way of, and third party tasting mode cost it is high,
It is whole to lay particular stress on, it is not suitable for the price evaluation of idle clothes yet.
Certainly, there are the value appraisal systems and method of various second-hand articles on the market, and more is exactly about two
The price evaluation and transaction system of handcart, however easy standardized commodity are different from, the clothing that leaves unused is due to a great variety, clothing valence
It is worth that evaluation criteria is fuzzy, and clothing is worth the factors such as wider range, there are certain difficulty on data acquisition and data processing, therefore mesh
It is preceding that the system for carrying out accurate valuation to clothes there is no to exist on the market.
Summary of the invention
The purpose of the present invention is to provide a kind of idle clothing valuation system and method based on data mining, the program are logical
The method that data mining is carried out to ten hundreds of transaction data of idle clothing is crossed, it is second-hand to commodity to obtain each product features
The feature weight of price obtains the mode of the second-hand calculation of price of idle clothing, gives both parties and meet marketability
Second-hand goods value expectation, value cognition of the balance both sides in transaction are poor.And then improve the experience of both parties' two-way trade.
In order to realize that any of the above goal of the invention, the present invention provide a kind of idle clothing valuation system based on data mining
System, comprising:
Data capture unit, obtains the characteristic group of idle clothing, and the characteristic in characteristic group includes but not
When being limited to correspond to the purchasing price of the idle clothing, brand classification, season, gender, the degree of wear, fabric, style and buying
Between;
Clothing database, the characteristic group that storing data acquiring unit obtains, and built-in correspondence leave unused clothing at least
One characteristic group;
Data preparation unit, arrange sort out clothing database in characteristic group, include at least: deletion lack two with
The characteristic group of upper characteristic, and sort out the characteristic group with same characteristic features data;
Data normalizing unit, the purchasing price of the gradient normalizing idle clothing obtain the gradient mark of the corresponding idle clothing
Quasi- purchasing price;
Feature abstraction unit quantifies non-quantized characteristic, obtains the feature weight of non-quantized characteristic, wherein non-amount
Change characteristic to include brand classification, season, gender, the degree of wear, fabric, style and buy the time;
Data linear regression unit, the feature weight and gradient standard purchase price that would sit idle for clothing substitute into linear regression mould
In type, decline formula in conjunction with gradient, the regression coefficient of character pair data be calculated, obtain valuation formula, valuation is public as follows:
Y=x (0.393x1+0.081x2+0.158x3+0.158x4+0.171x5+0.031x6), wherein x is purchase price
Lattice, x1 are brand, x2 is reduction ratio, x3 is gender, x4 is style, x5 is to buy time, x6 fabric.
According to another aspect of the present invention, the present invention provides a kind of idle clothing estimation method based on data mining, packet
Include following steps:
S1: the characteristic group of idle clothing is obtained, this feature data group includes but is not limited to following characteristics data: the spare time
It sets the buying price of clothing, brand classification, season, gender, the degree of wear, fabric, style, technique and buys the time;
S2: the multiple groups characteristic group in clothing database about the clothing that leaves unused is transferred in association;
S3: arranging and classification features data group, sorts out the characteristic group with same characteristic features data;
S4: buying price gradient normalizing is obtained at least one about the specific idle clothing by data normalizing and quantification treatment
Gradient standard purchase price, non-quantized characteristic quantifies to obtain at least feature for corresponding to each non-quantized characteristic power
Weight, wherein non-quantized characteristic includes brand classification, season, gender, the degree of wear, fabric, style and buys the time;
S5: the obtained feature weight of step S4 and gradient standard purchase price are substituted into linear regression model (LRM), in conjunction with ladder
Degree decline formula, is calculated the regression parameter of linear regression model (LRM), obtains valuation formula: y=x (0.393x1+0.081x2+
0.158x3+0.158x4+0.171x5+0.031x6), wherein x is buying price, x1 is brand, x2 is reduction ratio, x3 is property
Not, x4 is style, x5 is to buy time, x6 fabric.
Compared with the prior art, the present invention has the following beneficial effects:
1, it is excavated by the fetched data to idle clothing, to the buying price normalization of idle clothing, makes to count
According to normal distribution law is more in line with, data-handling efficiency when improving the subsequent quantization to non-quantized characteristic.
2, it is arranged by the recurrence to assessment data, it is second-hand for every idle clothing of product features quantization from big data
The feature weight of price improves the accuracy of the idle second-hand valuation price of clothing, further decrease predicted value and true value it
Between gap.
3, the real-time optimization of valuation formula updates, and can increase with the aggregate transaction data of time.By practical fetched data
It brings formula into as training input constantly to train, the accuracy of model formation also will gradually tend to be mature.
4, a kind of value appraisal system for idle clothes is provided, idle transaction underpants' article class nonstandardized technique is solved
Problem.Commodity value reference is provided to buy and sell double hairs.
Detailed description of the invention
Fig. 1 is that a kind of process of idle clothing estimation method based on data mining of an embodiment according to the present invention is shown
It is intended to.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art's every other embodiment obtained belong to what the present invention protected
Range.
It will be understood by those skilled in the art that in exposure of the invention, term " longitudinal direction ", " transverse direction ", "upper",
The orientation of the instructions such as "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom" "inner", "outside" or position are closed
System is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of description of the present invention and simplification of the description, without referring to
Show or imply that signified device or element must have a particular orientation, be constructed and operated in a specific orientation, therefore above-mentioned art
Language is not considered as limiting the invention.
It is understood that term " one " is interpreted as " at least one " or " one or more ", i.e., in one embodiment,
The quantity of one element can be one, and in a further embodiment, the quantity of the element can be it is multiple, term " one " is no
It can be interpreted as the limitation to quantity.
The present invention provides a kind of idle clothing valuation system based on data mining and corresponding estimation method, the system knot
Close data digging method, multiple linear regression analysis method, data normalization method, gradient descent method and sparse data processing side
The many algorithms such as method, establish it is a kind of can the commodity huge to this nonstandardized technique of clothing, trading volume carry out pretreatment valuation
System.
Fig. 1 gives the flow diagram of the idle clothing estimation method based on data mining, is data mining algorithm
Application of the middle multiple linear regression to idle clothing valuation, comprising the following steps: step 1: the recycling of second-hand idle clothing is obtained
Data --- step 2: store --- step 3: cleaning to data --- step 4: to the second-hand spare time to data collection permanence
Set the buying price data normalization processing of clothing --- step 5: the data after cleaning are obtained according to the value relevance of commodity
Feature weight --- the step 6 of each non-quantized characteristic: according to product features dimension referring to multiple linear regression model system
Determining model formation --- step 7: solving multiple linear regression equations using gradient descent algorithm ---, step 8: testing model misses
Difference.
The idle clothing valuation system based on data mining includes: data capture unit, which obtains
The characteristic group of specific idle clothing, the including but not limited to purchasing price of the idle clothing, brand classification, season, gender,
The degree of wear, fabric, style and buy the characteristics such as time.It is noted that the data capture unit can be
The idle clothing of intelligent terminal handles entrance, is also possible to market data collection.
Including clothing database, data capture unit described in the clothing database purchase obtain about the specific idle clothing
The characteristic group of object, and be built-in with database association inquiry system, can arrange and call about same specific idle clothing
Multiple groups characteristic group.That is, multiple groups characteristic group of the clothing database purchase about same specific idle clothing, and
Calling can be associated.
Including data preparation unit, it is special which arranges the multiple groups called out from the clothing database
Levy data group, the process arranged is as follows: 1. delete imperfect characteristic group: deleting the spy for lacking more than two characteristics
Data group is levied, for example, lacking fabric, technique or more features number into a certain characteristic group when data preparation unit is arranged
According to when, delete this group of characteristic group.2, the characteristic group with feature is clustered:, will be special according to the characteristic for the clothing that leaves unused
Sign data group is grouped, such as: the characteristic group of identical fabric is classified as identical fabric group, by the feature of same process
Data group is classified as same process group.
Including data normalizing unit, for standardizing the purchasing price of specific idle clothing, because buying price is to embody quotient
The core factor of product value, should not use the normalizing method that [- 1,1] is just being distributed very much, float up and down larger need additionally, due to price
Interal separation is carried out to it, the present invention carries out normalizing to buying price by the way of gradient normalizing, to setting Price Gradient
Section price cooperation dichotomy standardize each section Price factor by the way of (max price+min price)/2, obtain to
The gradient standard purchase price of a few specific idle clothing.
For example, when the serial buying price for being collected into specific idle clothing, there is 100,110,112,115,120,
121,129 etc., with 10 yuan for a gradient section, then the gradient standard purchase price in normalizing [100-110] gradient section are as follows:
105;Gradient standard purchase price in normalizing [111-120] gradient section are as follows: 116;In normalizing [121-130] gradient section
Gradient standard purchase price are as follows: 125.,
Including feature abstraction unit, for obtaining non-quantized characteristic to the feature weight of idle clothing valuation, especially
It is the commodity with multiple characteristics such to clothing, in the case where other characteristic values are identical or approximate situation.A certain feature
Linear dependence is presented with the price of clothing, wherein non-quantized characteristic include brand classification, season, gender, the degree of wear,
Fabric, style and buy the time.
That is, y=ax+b;Wherein y is second-hand expected price, and a feature weight, x is buying price, and b is disturbance term, is calculated
To feature weight, value is fallen between [0,1].In the case where other characteristic values are identical or approximate situation, a certain non-quantized characteristic
According to feature weight can be obtained by the formula, i.e., influence factor of the non-quantized characteristic to idle clothing valuation is converted to
The numerical value of dimension.
It when calculating, finds that other characteristic values are identical or the characteristic group of approximate data, substitutes into the second-hand valuation in market
Corresponding specific non-quantized feature is calculated using one-dimensional linear regression equation y=ax+b as basic skills in price and buying price
The feature weight of the series of data.
For example, same brand, same style, same fabric a clothing, only to men and women's classification to the shadow of price
Sound does regression analysis.Brand is the more than one piece housing of Hotwind, and fabric is pure cotton, and style is surplus, buys time phase difference not
It is more, it is autumn clothing, the price before depreciation is respectively 300 yuan, and 500 yuan, 100 yuan, market for men expectation valuation is 80 after depreciation
Member, for women is 250 yuan, and children's money is 10 yuan, then the corresponding feature weight of sex character is man 0.26, Ms 0.5, children
0.1。
In addition, it is noted that mean value filling can be used after the feature weight for obtaining non-quantized characteristic
Mode fills the characteristic group for losing a characteristic, that is, during arrangement, occurs lacking two following characteristics data
Characteristic group, the missing values can be filled with mean value after getting characteristic group, increase sample data volume.
Including data linear regression unit, which weighs the feature for obtaining specific idle clothing thereon
Weight and gradient standard purchase price substitute into linear regression model (LRM), obtain equation of linear regression.Specifically, data linear regression
Unit collects the data that obtain thereon, carries out preliminary analysis linear dependence, homoscedasticity, Multivariate Normal, low multiple to data
Synteny.Use multiple linear regression problem.It is solved using multiple linear regression.
Assuming that the given attribute dimensions vector with n attribute, in an embodiment of the present invention, n attribute representative influence
The non-quantized characteristic of idle clothing valuation:
Specifically, the attribute as the present invention influences clothes valuation has brand classification, season, gender, the degree of wear, face
Material, style, technique and the time is bought, corresponding x1 represents brand classified weight, and x2 represents season weight, and x3 represents gender power
Weight, x4 represent degree of wear weight, and x5 represents fabric weight, and x6 represents style weight, and x7 represents technique weight, and x8 representative is bought
Time weighting, x0 represent buying price.For clothes valuation, these characteristics will affect final second-hand valence
Value.
In addition each feature is one group of classification (example: seasonal characteristic group classification are as follows: spring, summer, autumn and winter), returns y by price
After=ax+b quantization, multiple corresponding characteristic quantification data groups are obtained.However influence of each characteristic to final price
Accounting is different, we using these characteristic groups as dependent variable bring multidimensional linear combination into predict, i.e. y=x0w1x1
+ x0x2w2+..x0xnwn, that is, y=x0 (w1x1+w2x2+ ... wnxn), herein, y are scheduled to last to price, and w indicates that correspondence is each non-
The regression coefficient of quantization characteristic data, i.e., the accounting of non-quantized characteristic, x are the spy of the non-quantized characteristic by quantization
Weight is levied, then is brought into after sample data obtains multiple regression equations, we can be write a Chinese character in simplified form into vector formIt calculates and obtains regression coefficient, wherein the standard for solving regression coefficient is the predicted value so that acquiring
Minimum with the gap of true value, we are dedicated to obtaining appropriateAnd b, so that the gap of the predicted value and true value that acquire is most
Small, function is most accurate.
Decline formula by gradient, in the hope of most suitable S:
To the lesser data of data volume, generally calculated using formula below:
Gradient descent method is used to the biggish data of data volume, gradient descent method herein and before one-variable linear regression
Gradient descent method is essentially identical, and only there are two the parameters that need to ask for one-variable linear regression, and have in multiple linear regression it is multiple to
Seek parameter.Remaining only needs to change derivative term.Finally obtained formula is as follows:
W0 is constant term
Sample buying price data, characteristic quantification data group and formula are passed through into python third party library TensorFlow/
Numpy/, which reads to complete to solve, is calculated each regression coefficient wj, and calculating instrument can be with multiple choices such as r, MATLAB, spass
Deng.
Finally obtain regression model:
Obtained valuation formula is as follows:
Y=x (0.393x1+0.081x2+0.158x3+0.158x4+0.171x5+0.031x6)
Including regression parameter authentication unit, regression parameter is verified by the following method:
1, a good multiple linear regression model should meet as far as possible:
(1) (numeric type) independent variable will be with the wired sexual intercourse of dependent variable;
(2) residual error is in normal distribution substantially;
(3) residual variance is basically unchanged (homoscedasticity);
(4) related independent between residual error (sample)
It can make comprehensive verification to the hypothesis of model of fit using gvlma () function of gvlma packet in R language thus, and right
Kurtosis, the degree of bias are verified.If detection is not by the way that linear hypothesis, normality, same to variance, independence four direction can be used also
Detection.
2, multicollinearity detects
The each independent variable of linear model in ideal should be linear independence, can if there are syntenies between independent variable
Reduce the accuracy of regression coefficient.It is generally measured with variance inflation factor VIF (Variance Inflation Factor) altogether
Linearly.
3, rejecting outliers
Including discrete group, High leverage Cases, strong and weak influence point.
In addition, the present invention provides a kind of idle clothing estimation method based on data mining, comprising the following steps:
S1: the characteristic group of specific idle clothing is obtained, this feature data group includes but is not limited to following characteristics data:
Purchasing price, brand classification, season, gender, the degree of wear, fabric, style and buy the time
S2: the multiple groups characteristic group in clothing database about the specific idle clothing is transferred in association;
S3: arranging and classification features data group, sorts out the characteristic group with same characteristic features data;
S4: buying price gradient normalizing is obtained at least one about this by data normalization and dimensional characteristics quantification treatment
The gradient standard purchase price of specific idle clothing, non-quantized characteristic dimension, which quantifies to obtain, corresponds to each non-quantized characteristic
According to an at least feature weight, wherein non-quantized characteristic includes brand classification, season, gender, the degree of wear, fabric, money
Formula and buy the time;
S5: the obtained feature weight of step S4 and gradient standard purchase price are substituted into linear regression model (LRM), in conjunction with ladder
Degree decline formula, is calculated the regression parameter of linear regression model (LRM), obtains valuation formula:
Valuation formula: y=x (0.393x1+0.081x2+0.158x3+0.158x4+0.171x5+0.031x6) is obtained,
Middle x is buying price, x1 is brand, x2 is reduction ratio, x3 is gender, x4 is style, x5 is to buy time, x6 fabric.
Certainly, in the method include step S6: the verifying of linear regression model (LRM) model, referring to recurrence Verification unit portion
The introduction divided carries out the verifying of regression parameter.
After step S4 gets feature weight, is filled by the way of mean value filling and lack two following characteristics data
Characteristic group, this feature data group are back in clothing database, make full use of data.
Specifically, specifically to adjust process as follows by the step S3: 1. delete imperfect characteristic groups: deleting missing
The characteristic group of more than two characteristics, for example, lacking fabric, technique or more into a certain characteristic group when arranging
When multiple features data, this group of characteristic group is deleted.2, the characteristic group with feature is clustered: the characteristic according to the clothing that leaves unused
According to, characteristic group is grouped, such as: the characteristic group of identical fabric is classified as identical fabric group, by identical work
The characteristic group of skill is classified as same process group.
Step S4 includes buying price data normalizing and non-quantized characteristic normalizing:
The method of buying price data normalizing is as follows: using (max to the price cooperation dichotomy in setting Price Gradient section
Price+min price)/2 mode standardizes each section Price factor, obtain the gradient standard purchase of at least one specific idle clothing
Enter price;
Here non-quantized characteristic refers herein to remove other characteristics other than buying price, needs to obtain
Feature weight of the negated quantization characteristic data to idle clothing valuation, that is, y=ax+b;Wherein y is second-hand expected price, and a is back
Return parameter, x is buying price, and b is disturbance term.It finds that other characteristic values are identical or the characteristic group of approximate data, substitutes into
The regression parameter of the series of corresponding specific non-quantized characteristic is calculated in the second-hand valuation price in market and buying price.
Equation of linear regression in S5 are as follows: y=x0w1x1+x0x2w2+..x0xnwn, that is, y=x0 (w1x1+w2x2+ ...
Wnxn), wherein y is scheduled to last to price, and w indicates the regression coefficient of corresponding each non-quantized characteristic, i.e., non-quantized characteristic
Accounting, substitute into gradient decline formula, obtain predicted value and the smallest regression parameter of true value gap.
Wherein, the method declined by gradient, in the hope of most suitable S:
To the lesser data of data volume, calculated using formula below:
Gradient descent method is used to the biggish data of data volume, gradient descent method herein and before one-variable linear regression
Gradient descent method is essentially identical, and only there are two the parameters that need to ask for one-variable linear regression, and have in multiple linear regression it is multiple to
Seek parameter.Remaining only needs to change derivative term.Finally obtained formula is as follows:
W0 is constant term;
Calculation and tool can choose to be calculated respectively for python third party library TensorFlow/numpy/ completion
Regression coefficient, calculating instrument can be with multiple choices such as r, MATLAB, spass etc..
The present invention is not limited to above-mentioned preferred forms, anyone can show that other are various under the inspiration of the present invention
The product of form, however, make any variation in its shape or structure, it is all that there is skill identical or similar to the present application
Art scheme, is within the scope of the present invention.
Claims (10)
1. a kind of idle clothing valuation system based on data mining characterized by comprising
Data capture unit, obtains the characteristic group of idle clothing, and the characteristic in characteristic group includes but is not limited to
The purchasing price of the corresponding idle clothing, brand classification, season, gender, the degree of wear, fabric, style and buy the time;
At least the one of the idle clothing of clothing database, the characteristic group that storing data acquiring unit obtains, and built-in correspondence is special
Levy data group;
Data preparation unit arranges the characteristic group sorted out in clothing database, includes at least: deletion lacks more than two spies
The characteristic group of data is levied, and sorts out the characteristic group with same characteristic features data;
Data normalizing unit, the purchasing price of the gradient normalizing idle clothing obtain the gradient standard purchase of the corresponding idle clothing
Buying price lattice;
Feature abstraction unit quantifies non-quantized characteristic, obtains the feature weight of non-quantized characteristic, wherein non-quantized spy
Sign data include brand classification, season, gender, the degree of wear, fabric, style and buy the time;
Data linear regression unit, the feature weight and gradient standard purchase price that would sit idle for clothing substitute into linear regression model (LRM)
In, decline formula in conjunction with gradient, the regression coefficient of character pair data be calculated, obtain valuation formula, valuation is public as follows:
Y=x (0.393x1+0.081x2+0.158x3+0.158x4+0.171x5+0.031x6), wherein x is buying price, x1
For brand, x2 be reduction ratio, x3 is gender, x4 be style, x5 be buy the time, x6 is fabric.
2. the idle clothing valuation system according to claim 1 based on data mining, which is characterized in that linear regression mould
Type is y=x0 (w1x1+w2x2+ ... wnxn), and y is known expectation price, and w indicates returning for corresponding each non-quantized characteristic
Return coefficient, x is the feature weight of the non-quantized characteristic by quantization using gradient descent algorithm, calculates acquisition and returns system
Number, wherein the standard for solving regression coefficient is so that the gap of the predicted value and true value that acquire is minimum.
3. the idle clothing valuation system according to claim 2 based on data mining, which is characterized in that when data volume phase
To it is little when, gradient decline formula it is as follows:When data volume is relatively large, under gradient
It is as follows that formula drops:
w0It is constant term
4. according to claim 1 to 3 any idle clothing valuation systems based on data mining, which is characterized in that special
The method for levying the non-quantized characteristic of quantization of abstraction unit is as follows: being based on equation of linear regression y=ax+b, selects other features
Under data dimension is identical or approximate situation, the different clothing price data group of a certain non-quantized characteristic, using y=ax+b;
Wherein y is second-hand expected price, and a is characterized weight, and x is original purchase money lattice, and b is disturbance term, and feature weight, value is calculated
It falls between [0,1].
5. according to claim 1 to 3 any idle clothing valuation systems based on data mining, which is characterized in that number
It is as follows according to the price data normalizing method of normalizing unit: (max valence is used to the price cooperation dichotomy in setting Price Gradient section
Lattice+min price)/2 mode standardizes each section Price factor, obtain corresponding idle clothing pair gradient standard buy
Price.
6. according to claim 1 to 3 any idle clothing valuation systems based on data mining, which is characterized in that
After the feature weight for obtaining non-quantized characteristic, the characteristic for losing a characteristic is filled by the way of mean value filling
According to group;Regression coefficient authentication unit is also comprised, regression coefficient is verified.
7. a kind of idle clothing estimation method based on data mining, which comprises the following steps:
S1: the characteristic group of idle clothing is obtained, this feature data group includes but is not limited to following characteristics data: the idle clothing
The buying price of object, brand classification, season, gender, the degree of wear, fabric, style and buy the time;
S2: the multiple groups characteristic group in clothing database about the clothing that leaves unused is transferred in association;
S3: arranging and classification features data group, sorts out the characteristic group with same characteristic features data;
S4: buying price gradient normalizing is obtained at least one ladder about the specific idle clothing by data normalizing and quantification treatment
Standard purchase price is spent, non-quantized characteristic quantifies to obtain at least feature weight for corresponding to each non-quantized characteristic,
Wherein non-quantized characteristic includes brand classification, season, gender, the degree of wear, fabric, style and buys the time;
S5: the obtained feature weight of step S4 and gradient standard purchase price are substituted into linear regression model (LRM), in conjunction under gradient
Formula is dropped, the regression coefficient of linear regression model (LRM) is calculated, obtains valuation formula: y=x (0.393x1+0.081x2+
0.158x3+0.158x4+0.171x5+0.031x6)。
8. the idle clothing estimation method according to claim 7 based on data mining, which is characterized in that described linear time
Returning model is y=x0 (w1x1+w2x2+ ... wnxn), and y is known expectation price, and w indicates corresponding each non-quantized characteristic
Regression coefficient, x is the feature weight of the non-quantized characteristic by quantization using gradient descent algorithm, calculates to obtain and returns
Coefficient, wherein the standard for solving regression coefficient is so that the gap of the predicted value and true value that acquire is minimum.
9. the idle clothing estimation method according to claim 8 based on data mining, which is characterized in that when data volume phase
To it is little when, gradient decline formula it is as follows:
When data volume is relatively large, it is as follows that gradient declines formula:
ω0It is constant term
10. the idle clothing estimation method according to claim 6 based on data mining, which is characterized in that the step
In S4, the price cooperation dichotomy in setting Price Gradient section is standardized respectively by the way of (max price+min price)/2
Section Price factor obtains an at least gradient standard purchase price;Based on equation of linear regression y=ax+b, other features are selected
Under data dimension is identical or approximate situation, the different clothing price data group of a certain non-quantized characteristic, using y=ax+b;
Wherein y is second-hand expected price, and a is characterized weight, and x is original purchase money lattice, and b is disturbance term, and feature weight, value is calculated
It falls between [0,1].
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910283617.1A CN110046934A (en) | 2019-04-10 | 2019-04-10 | A kind of idle clothing valuation system and method based on data mining |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910283617.1A CN110046934A (en) | 2019-04-10 | 2019-04-10 | A kind of idle clothing valuation system and method based on data mining |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110046934A true CN110046934A (en) | 2019-07-23 |
Family
ID=67276563
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910283617.1A Pending CN110046934A (en) | 2019-04-10 | 2019-04-10 | A kind of idle clothing valuation system and method based on data mining |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110046934A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113052657A (en) * | 2021-04-01 | 2021-06-29 | 武汉畅来电子商务有限公司 | Online commodity intelligent matching recommendation method based on user behavior analysis and cloud server |
CN113239323A (en) * | 2021-03-24 | 2021-08-10 | 深圳市雅昌艺术网股份有限公司 | Artwork valuation system and valuation method based on transaction data |
CN113327127A (en) * | 2021-05-26 | 2021-08-31 | 深圳闪回科技有限公司 | Automatic measuring, calculating and price-adjusting system |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103279879A (en) * | 2013-05-29 | 2013-09-04 | 浙江搜车客网络技术有限公司 | Method for online valuation of used cars |
CN106157090A (en) * | 2016-06-28 | 2016-11-23 | 江苏车置宝信息科技股份有限公司 | The online second-hand car fare appraisal procedure realized based on big data |
CN106651462A (en) * | 2016-12-29 | 2017-05-10 | 上海为秀互联网科技有限公司 | Evaluation method for second-hand device |
CN107633435A (en) * | 2017-10-09 | 2018-01-26 | 广州市万表科技股份有限公司 | A kind of online estimation method of second-hand goods and system |
CN108564415A (en) * | 2018-04-25 | 2018-09-21 | 济南浪潮高新科技投资发展有限公司 | A kind of method of intelligent predicting used car price |
CN109583926A (en) * | 2017-09-29 | 2019-04-05 | 优估(上海)信息科技有限公司 | A kind of used car pricing method apparatus and system |
-
2019
- 2019-04-10 CN CN201910283617.1A patent/CN110046934A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103279879A (en) * | 2013-05-29 | 2013-09-04 | 浙江搜车客网络技术有限公司 | Method for online valuation of used cars |
CN106157090A (en) * | 2016-06-28 | 2016-11-23 | 江苏车置宝信息科技股份有限公司 | The online second-hand car fare appraisal procedure realized based on big data |
CN106651462A (en) * | 2016-12-29 | 2017-05-10 | 上海为秀互联网科技有限公司 | Evaluation method for second-hand device |
CN109583926A (en) * | 2017-09-29 | 2019-04-05 | 优估(上海)信息科技有限公司 | A kind of used car pricing method apparatus and system |
CN107633435A (en) * | 2017-10-09 | 2018-01-26 | 广州市万表科技股份有限公司 | A kind of online estimation method of second-hand goods and system |
CN108564415A (en) * | 2018-04-25 | 2018-09-21 | 济南浪潮高新科技投资发展有限公司 | A kind of method of intelligent predicting used car price |
Non-Patent Citations (1)
Title |
---|
孙娅楠: "梯度下降法在机器学习中的应用", 《中国优秀硕士学位论文全文数据库(信息科技辑)》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113239323A (en) * | 2021-03-24 | 2021-08-10 | 深圳市雅昌艺术网股份有限公司 | Artwork valuation system and valuation method based on transaction data |
CN113052657A (en) * | 2021-04-01 | 2021-06-29 | 武汉畅来电子商务有限公司 | Online commodity intelligent matching recommendation method based on user behavior analysis and cloud server |
CN113052657B (en) * | 2021-04-01 | 2021-12-21 | 深圳市沃客非凡科技有限公司 | Online commodity intelligent matching recommendation method based on user behavior analysis and cloud server |
CN113327127A (en) * | 2021-05-26 | 2021-08-31 | 深圳闪回科技有限公司 | Automatic measuring, calculating and price-adjusting system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Bikker | An international trade flow model with substitution: an extension of the gravity model | |
CN110046934A (en) | A kind of idle clothing valuation system and method based on data mining | |
Buhmann et al. | Equivalence scales, well‐being, inequality, and poverty: sensitivity estimates across ten countries using the Luxembourg Income Study (LIS) database | |
US7346421B2 (en) | Production of made to order clothing | |
Cho | Tourism forecasting and its relationship with leading economic indicators | |
CN108256691A (en) | Refund Probabilistic Prediction Model construction method and device | |
CN108733631A (en) | A kind of data assessment method, apparatus, terminal device and storage medium | |
CN107451894A (en) | Data processing method, device and computer-readable recording medium | |
CN107230113A (en) | A kind of house property appraisal procedure of multi-model fusion | |
CN107437203A (en) | Information-pushing method, device, electronic installation and computer-readable medium | |
CN106485562A (en) | A kind of commodity information recommendation method based on user's history behavior and system | |
CN107832964A (en) | Bank client relation loop analysis method and system | |
CN107358505A (en) | A kind of size for purchase clothing online recommends method and system | |
CN107167565A (en) | A kind of Table Grape method for evaluating quality and system | |
CN107180160A (en) | Public bicycles consumer loyalty degree based on SEM models determines method | |
CN109559199A (en) | A kind of product methods of sampling of reflection online shopping customer quality experience | |
CN109472662A (en) | A kind of commodity selling method and device, electronic equipment and storage medium | |
CN111680910A (en) | Tobacco application range positioning method based on cigarette formula efficacy | |
CN114782173A (en) | Credit risk assessment method and device, electronic equipment and storage medium | |
CN112506997B (en) | Big data user mining method based on cross-border e-commerce platform | |
CN112967108B (en) | BP-ANN-based power retail package search ordering simulation system and method | |
CN110990782B (en) | Cigarette sensory quality evaluation method based on weighted hesitation fuzzy power aggregation operator | |
CN107392679A (en) | A kind of data processing method of intellectual property value analysis identification, apparatus and system | |
EP1341427B1 (en) | Production of made to order clothing | |
CN116188039A (en) | Intelligent recommendation method and system for suppliers |
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 | ||
CB02 | Change of applicant information |
Address after: 310006 room 1101, building 2, Xingcheng development building, Xiacheng District, Hangzhou City, Zhejiang Province Applicant after: Zhejiang Gubao Technology Co., Ltd Address before: 310006 room 1101, building 2, Xingcheng development building, Xiacheng District, Hangzhou City, Zhejiang Province Applicant before: Hangzhou Gubao e-commerce Co., Ltd |
|
CB02 | Change of applicant information |