CN108319611A - The methods of sampling and sampling apparatus - Google Patents
The methods of sampling and sampling apparatus Download PDFInfo
- Publication number
- CN108319611A CN108319611A CN201710035012.1A CN201710035012A CN108319611A CN 108319611 A CN108319611 A CN 108319611A CN 201710035012 A CN201710035012 A CN 201710035012A CN 108319611 A CN108319611 A CN 108319611A
- Authority
- CN
- China
- Prior art keywords
- amount
- sampling
- sublayer
- sample set
- sample
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/907—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
Abstract
The present invention proposes a kind of methods of sampling and a kind of sampling apparatus, wherein the methods of sampling includes：According to specified sublayer attribute, layered shaping is carried out to experiment sample set and check sample set respectively, to obtain the second amount of sampling of the first amount of sampling of each sublayer of experiment sample set and each sublayer of check sample set；When the second amount of sampling of each sublayer is not more than the first amount of sampling, or, when first amount of sampling it is expected layering amount no more than control, the first object proportionate relationship of the first specified requirements of selection satisfaction in the proportionate relationship from the first amount of sampling of each sublayer with the second amount of sampling；According to the first amount of sampling of first object proportionate relationship and each sublayer, the third amount of sampling of each sublayer is obtained；Processing is sampled to experiment sample set and check sample set according to third amount of sampling.Technical solution through the invention can extract more A representative Sample, to improve the validity that result is investigated in sampling.
Description
【Technical field】
The present invention relates to technical field of data processing more particularly to a kind of methods of sampling and a kind of sampling apparatus.
【Background technology】
Currently, when being sampled investigation to two user groups, there are two types of usual ways for tool：
First, it gives two user groups and respectively needs the sample size extracted, to the two users by the way of random sampling
Group carries out balanced sampling sampling.
Second, random sampling can be carried out according to a few of numerical account number to the user group with numerical account number system,
For example, the front three for extracting numerical account number be 839, it is intermediate three be random number, it is rear two be 07 user as sample, then can
Generate 1000 random users that the centre three of numerical account number is 000 to 999.
However, during actually sampling is investigated, other dimension differences other than dimension to be investigated are smaller, treat investigation dimension
The sampling comparison result of degree just more accurate more closing to reality, and the whole difference of two user groups is often very big, user is in year
A variety of dimensions such as age, gender, location having differences property.Therefore, using the random sampling of first way, phase can not be extracted into
Itself there is sample representative enough for user group, correspondingly, sampling comparison result is also unreliable.And second
Mode only ensure that the uniformity that user is numerically distributed in account, is still substantially random sampling, therefore can not also be extracted into
Itself there is sample representative enough relative to user group.
Therefore, more A representative Sample how is extracted, to further increase the reliability of sampling comparison result, becomes mesh
Preceding technical problem urgently to be resolved hurrily.
【Invention content】
An embodiment of the present invention provides a kind of methods of samplings and a kind of sampling apparatus, it is intended to solve in the related technology because extracting
Sample do not have the representative and insecure technical problem of comparison result that causes to draw a lottery, more representative sample can be extracted
This, further increases the reliability of sampling comparison result.
In a first aspect, an embodiment of the present invention provides a kind of methods of samplings, including：According to specified sublayer attribute, respectively
Layered shaping is carried out to experiment sample set and check sample set, is taken out with obtaining the first of each sublayer of experiment sample set
Second amount of sampling of each sublayer of sample amount and check sample set；When the second amount of sampling of each sublayer is taken out no more than first
Sample amount, alternatively, when the first amount of sampling it is expected layering amount no more than control, from the first amount of sampling and the second amount of sampling of each sublayer
Proportionate relationship in selection meet the first specified requirements first object proportionate relationship；According to the first object proportionate relationship and
First amount of sampling of each sublayer, obtains the third amount of sampling of each sublayer；According to third amount of sampling to the experiment sample collection
It closes and the check sample set is sampled processing.
In the above embodiment of the present invention, optionally, further include：When the second amount of sampling of each sublayer is more than the first sampling
Amount, and the first amount of sampling be more than control it is expected layering amount when, based on control it is expected layering amount to the experiment sample set and institute
It states check sample set and is sampled processing.
In the above embodiment of the present invention, optionally, from the ratio of the first amount of sampling and the second amount of sampling of each sublayer
After selection meets the first object proportionate relationship of the first specified requirements in relationship, further include：Judge the first object ratio
Whether relationship meets the second specified requirements；If the first object proportionate relationship meets the second specified requirements, execute step according to
First amount of sampling of the first object proportionate relationship and each sublayer obtains the third amount of sampling of each sublayer；If described
One target proportion relationship is unsatisfactory for the second specified requirements, then from other proportionate relationships in addition to the first object proportionate relationship
Middle selection meets the second target proportion relationship of the first specified requirements, according to the second target proportion relationship and each sublayer
First amount of sampling obtains the third amount of sampling of each sublayer.
In the above embodiment of the present invention, optionally, first specified requirements is the second amount of sampling and the first amount of sampling
Ratio it is minimum；Or the ratio maximum that first specified requirements is the first amount of sampling and the second amount of sampling.
In the above embodiment of the present invention, optionally, second specified requirements is the first object proportionate relationship and 0
Difference be more than specified threshold；Or second specified requirements is equal to finger for the difference of the first object proportionate relationship and 0
Determine threshold value.
In the above embodiment of the present invention, optionally, according to specified sublayer attribute, respectively to experiment sample set and right
Layered shaping is carried out according to sample set, including：The sample with the sublayer attribute is extracted from the experiment sample set, it should
The total amount of sample is as first amount of sampling；And it is extracted with the sublayer attribute from the check sample set
Sample, the total amount of the sample is as second amount of sampling.
Second aspect, an embodiment of the present invention provides a kind of sampling apparatus, including：Stratified sampling amount acquiring unit, according to
Specified sublayer attribute carries out layered shaping, to obtain experiment sample collection to experiment sample set and check sample set respectively
Second amount of sampling of the first amount of sampling of each sublayer closed and each sublayer of check sample set；First selecting unit,
When each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling no more than control it is expected layering amount when,
Meet the first object of the first specified requirements from selection in the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
Proportionate relationship；Coordinate amount of sampling acquiring unit to be obtained according to the first amount of sampling of the first object proportionate relationship and each sublayer
To the third amount of sampling of each sublayer；Sample process unit, according to third amount of sampling to the experiment sample set and described right
It is sampled processing according to sample set.
In the above embodiment of the present invention, optionally, the sample process unit is additionally operable to：Second when each sublayer takes out
Sample amount be more than the first amount of sampling, and the first amount of sampling be more than control it is expected layering amount when, based on control it is expected layering amount to described
Experiment sample set and the check sample set are sampled processing.
In the above embodiment of the present invention, optionally, further include：Judging unit is full in first selecting unit selection
After the first object proportionate relationship of the first specified requirements of foot, it is specified to judge whether the first object proportionate relationship meets second
Condition；The coordination amount of sampling acquiring unit is specifically used for：If the first object proportionate relationship meets the second specified requirements, hold
Row step obtains the third sampling of each sublayer according to the first amount of sampling of the first object proportionate relationship and each sublayer
Amount, and if the first object proportionate relationship be unsatisfactory for the second specified requirements, from except the first object proportionate relationship with
Selection meets the second target proportion relationship of the first specified requirements in other outer proportionate relationships, according to second target proportion
First amount of sampling of relationship and each sublayer obtains the third amount of sampling of each sublayer.
In the above embodiment of the present invention, optionally, first specified requirements is the second amount of sampling and the first amount of sampling
Ratio it is minimum；Or the ratio maximum that first specified requirements is the first amount of sampling and the second amount of sampling.
In the above embodiment of the present invention, optionally, second specified requirements is the first object proportionate relationship and 0
Difference be more than specified threshold；Or second specified requirements is equal to finger for the difference of the first object proportionate relationship and 0
Determine threshold value.
In the above embodiment of the present invention, optionally, the stratified sampling amount acquiring unit is specifically used for：From the experiment
In sample set extract with the sublayer attribute sample, the total amount of the sample as first amount of sampling, and, from institute
It states and extracts the sample with the sublayer attribute in check sample set, the total amount of the sample is as second amount of sampling.
By above technical scheme, in the related technology because the sample of extraction do not have it is representative due to cause prize drawing to be compared
As a result insecure technical problem can first divide experiment sample set and check sample set according to a variety of sublayer attributes
Layer sampling, respectively obtains multiple sublayers, the son in the check sample set for keeping each sublayer of experiment sample set corresponding
Other attributes other than the attribute to be investigated of layer are consistent, in other words, that is, make each sublayer of experiment sample set
The sample distribution in sublayer in corresponding check sample set is consistent, the sample comparativity both to improve, with
Increase the reliability of the two comparison result.
Then, for each sublayer, if its corresponding control it is expected that layering amount is less than it in experiment sample set
The first amount of sampling, also, first amount of sampling be less than its in the second amount of sampling of check sample set, explanation can be each
Sublayer is drawn into number and reaches the sample that layering amount it is expected in its corresponding control.
Otherwise, that is, when each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling is little
When layering amount it is expected in control, illustrate that the first amount of sampling of the sublayer or the second amount of sampling are not achieved control and it is expected layering amount.
In this case, first object ratio can be selected from the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
The corresponding control of all sublayers it is expected layering amount scaled down, first object by relationship by the first object proportionate relationship
Proportionate relationship should meet the first specified requirements, can just obtain the third sampling for being both less than the first amount of sampling again smaller than the second amount of sampling
Amount finally carries out balanced sampling to experiment sample set and check sample set by third amount of sampling, ensure that balanced sample
The consistency of amount.
To sum up, technical scheme of the present invention is capable of providing a kind of new sample mode, sample drawn distribution situation and sample
The consistent balanced sample of total amount, such balanced sample have more comparativity, are compared and examined using such balanced sample
It examines, the reliability of sampling comparison result can be further increased, promote the objectivity and science of Experimental comparison.
【Description of the drawings】
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without having to pay creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 shows the flow chart for the methods of sampling that one embodiment of the present of invention provides；
Fig. 2 shows the flow charts for the methods of sampling that an alternative embodiment of the invention provides；
Fig. 3 shows the overall schematic for the data sampling process that one embodiment of the present of invention provides；
Fig. 4 shows the block diagram for the sampling apparatus that one embodiment of the present of invention provides；
Fig. 5 shows the block diagram for the terminal that one embodiment of the present of invention provides；
Fig. 6 shows the block diagram for the terminal that an alternative embodiment of the invention provides.
【Specific implementation mode】
For a better understanding of the technical solution of the present invention, being retouched in detail to the embodiment of the present invention below in conjunction with the accompanying drawings
It states.
It will be appreciated that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, those of ordinary skill in the art obtained without creative efforts it is all its
Its embodiment, shall fall within the protection scope of the present invention.
The term used in embodiments of the present invention is the purpose only merely for description specific embodiment, is not intended to be limiting
The present invention.In the embodiment of the present invention and "an" of singulative used in the attached claims, " described " and "the"
It is also intended to including most forms, unless context clearly shows that other meanings.
Fig. 1 shows the flow chart for the methods of sampling that one embodiment of the present of invention provides.
As shown in Figure 1, the methods of sampling that one embodiment of the present of invention provides, including：
Step 102, according to specified sublayer attribute, experiment sample set and check sample set are carried out at layering respectively
Reason is taken out with obtain the first amount of sampling of each sublayer of experiment sample set and each sublayer of check sample set second
Sample amount.
Layered shaping is referred to experiment sample set and check sample set, according to a variety of sublayer attributes and per seed
The segments of layer attribute, is divided into multiple sublayers, principle can be regarded as：
First, a variety of sublayer attributes of experiment sample set and check sample set and the segmentation per seed layer attribute are determined
Number, for example, when determining to using the crowd of specific APP to be sampled investigation, it may be determined that two seed layer attribute of age and gender,
Next it may be determined that the segments at age be 3, be divided into under-18s, 19 to 35 years old, 35 years old or more three sections, other point of certainty
Hop count is 2, is divided into two sections of man and female.
Then, experiment sample set and check sample set are layered according to the segmentation of the first seed layer attribute, are obtained
To multiple subclass, each subclass is layered according still further to second of sublayer attribute, and so on, until having taken turns all sons
Layer attribute, obtains final multiple sublayers.In this way, the check sample collection for just making each sublayer of experiment sample set corresponding
Other attributes other than the attribute to be investigated of sublayer in conjunction are consistent, in other words, that is, make experiment sample set
The sample distribution in sublayer in the corresponding check sample set of each sublayer is consistent, and the sample both to improve can
Than property, to increase the reliability of the two comparison result.
For example, 4 sublayer attributes can be taken, first sublayer attribute segments number is A, and second sublayer attribute segments number is
G, third sublayer attribute segments number are L, and the 4th sublayer attribute segments number is H, and it is M, total size to enable sample drawn sum
For N, then final layering sum is A × G × L × H.
In addition, above-mentioned principle is only to help to understand, in actual treatment, can often be directly obtained by predetermined hierarchical function
Layered shaping as a result, such as interior graceful distribution.
Step 104, when the second amount of sampling of each sublayer be not more than the first amount of sampling, alternatively, the first amount of sampling is not more than
When layering amount it is expected in control, in the proportionate relationship from the first amount of sampling of each sublayer with the second amount of sampling selection meet the first finger
The first object proportionate relationship of fixed condition.
Wherein, often have preset expectation check sample total amount, the preset expectation check sample total when sampling is compared
Amount is configured by user according to actual demand, can also be estimated to obtain according to the total amount of experiment sample set by system,
For example, system is preset as it is expected check sample total amount automatically by the 1% of experiment sample set.It is total with preset expectation check sample
Amount is multiplied by accounting of first amount of sampling in experiment sample set of any sublayer, can obtain the corresponding control of the sublayer and it is expected layering
Amount.
When each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling no more than control it is expected
When layering amount, illustrate that the first amount of sampling of the sublayer or the second amount of sampling are not achieved control and it is expected layering amount, at this time, it may be necessary to foundation
The first object proportionate relationship for meeting the first specified requirements carries out following processing.
Step 106, according to the first amount of sampling of first object proportionate relationship and each sublayer, the third of each sublayer is obtained
Amount of sampling.
First object ratio can be selected from the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
The corresponding control of all sublayers it is expected layering amount scaled down, first object by relationship by the first object proportionate relationship
Proportionate relationship should meet the first specified requirements, can just obtain the third sampling for being both less than the first amount of sampling again smaller than the second amount of sampling
Amount finally carries out balanced sampling to experiment sample set and check sample set by third amount of sampling, ensure that balanced sample
The consistency of amount.
Step 108, processing is sampled to experiment sample set and check sample set according to third amount of sampling.
Wherein, you need to add is that, the first specified requirements is the ratio minimum of the second amount of sampling and the first amount of sampling, or
Person is the ratio maximum of the first amount of sampling and the second amount of sampling.
In actual scene the first amount of sampling of experiment sample set commonly greater than check sample set the second amount of sampling,
The ratio of second amount of sampling and the first amount of sampling is smaller, that is, when the ratio of the first amount of sampling and the second amount of sampling is bigger, says
Both bright gap is bigger.Therefore, may be selected the second amount of sampling and the first amount of sampling ratio it is minimum or for the first amount of sampling and
The maximum first object proportionate relationship of ratio of second amount of sampling will be reduced according to desired layering amount, obtain third amount of sampling,
Third amount of sampling had both been less than the first amount of sampling again smaller than the second amount of sampling, in this way, using third amount of sampling to experiment sample set
It is sampled processing with check sample set, you can obtain the consistent sampling samples of quantity.
To sum up, a kind of new sample mode is present embodiments provided, it being capable of sample drawn distribution situation and sample total one
The balanced sample of cause, such balanced sample have more comparativity, are compared and investigated using such balanced sample, can be into
One step increases the reliability of sampling comparison result, promotes the objectivity and science of Experimental comparison.
Fig. 2 shows the flow charts for the methods of sampling that an alternative embodiment of the invention provides.
As shown in Fig. 2, the methods of sampling that an alternative embodiment of the invention provides, including：
Step 202, according to specified sublayer attribute, experiment sample set and check sample set are carried out at layering respectively
Reason is taken out with obtain the first amount of sampling of each sublayer of experiment sample set and each sublayer of check sample set second
Sample amount.
In this way, the sample in sublayer in the check sample set that may make each sublayer of experiment sample set corresponding
This distribution is consistent, the sample comparativity both to improve, to increase the reliability of the two comparison result.
Step 204, judge whether each sublayer the second amount of sampling be more than the first amount of sampling, and the first amount of sampling be more than pair
206 are entered step, otherwise, enters step 208 when judging result is to be according to desired layering amount.
Step 206, it is expected that layering amount is sampled processing to experiment sample set and check sample set based on control, tie
Shu Jincheng.
For each sublayer, if its corresponding control it is expected that layering amount is less than it the first of experiment sample set
Amount of sampling, also, first amount of sampling is less than it in the second amount of sampling of check sample set, and explanation can take out in each sublayer
It gets number and reaches the sample of its corresponding control expectation layering amount, therefore experiment sample set and check sample set can be taken out
Control is taken it is expected the sample of layering amount.
Step 208, selection meets the first finger from the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
The first object proportionate relationship of fixed condition.Wherein, the first specified requirements is the ratio minimum of the second amount of sampling and the first amount of sampling；
Or the ratio maximum that first specified requirements is the first amount of sampling and the second amount of sampling.
When each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling no more than control it is expected
When layering amount, illustrate that the first amount of sampling of the sublayer or the second amount of sampling are not achieved control and it is expected layering amount.And in actual scene
First amount of sampling of middle experiment sample set commonly greater than check sample set the second amount of sampling, the second amount of sampling and first take out
The ratio of sample amount is smaller, that is, when the ratio of the first amount of sampling and the second amount of sampling is bigger, both illustrates that gap is bigger.
Therefore, the ratio minimum of the second amount of sampling and the first amount of sampling may be selected or sample for the first amount of sampling and second
The maximum first object proportionate relationship of ratio of amount will be reduced according to desired layering amount, obtain third amount of sampling, third sampling
Amount had both been less than the first amount of sampling again smaller than the second amount of sampling, in this way, using third amount of sampling to experiment sample set and control sample
This set is sampled processing, you can obtains the consistent sampling samples of quantity.
Step 210, judge whether first object proportionate relationship meets the second specified requirements, when judging result is to be, into
Enter step 212, otherwise, enters step 214.Wherein, the second specified requirements be the first object proportionate relationship with 0 difference it is big
In specified threshold；Or second specified requirements is equal to specified threshold for the difference of the first object proportionate relationship and 0.
In actual scene, it is possible to can encounter sample number of the user group under some sublayer be 0 the case where, i.e., first
The case where target proportion relationship is 0, can not take out any sample at this time, therefore in order to coordinate this special circumstances, it can be to first
Target proportion relationship carries out a step detection, and specially treated is carried out to be directed to the case where first object proportionate relationship is 0.
Step 212, according to the first amount of sampling of first object proportionate relationship and each sublayer, the third of each sublayer is obtained
Amount of sampling.At this point, first object proportionate relationship is not 0, it can be corresponding right by all sublayers by the first object proportionate relationship
According to desired layering amount scaled down, third amount of sampling is obtained.
Step 214, selection meets the first specified requirements from other proportionate relationships in addition to first object proportionate relationship
The second target proportion relationship each sublayer is obtained according to the first amount of sampling of the second target proportion relationship and each sublayer
Third amount of sampling.
At this point, first object proportionate relationship is 0, the second target proportion for meeting the first specified requirements can be further selected
Relationship passes through for example, minimum other than 0 in multiple proportionate relationships one can be selected as the second target proportion relationship
The corresponding control of all sublayers it is expected layering amount scaled down by the second target proportion relationship, obtains third amount of sampling.
Step 216, processing is sampled to experiment sample set and check sample set according to third amount of sampling, so as to
Enough sample drawn distribution situations balanced sample consistent with sample total, further increases the reliability of sampling comparison result, carries
Rise the objectivity and science of Experimental comparison.
Fig. 3 shows the overall schematic for the data sampling process that one embodiment of the present of invention provides.
In general, experiment sample set be much smaller than check sample set, for example, advertisement launch in exposure group scale much
Less than non-exposed group of scale in deep bid any active ues.The total amount of experiment sample set is N, and amount of sampling allows to estimate according to experimental subjects
The maximum variance of metering is V, population variance S2, as shown in figure 3, then the calculation formula of the amount of sampling n of experiment sample set is：
However, under normal circumstances, given it is expected amount of sampling n of the check sample total amount commonly greater than experiment sample set,
Therefore, after determining amount of sampling n, if detecting that amount of sampling n is less than given expectation check sample total amount, it is necessary to reality
It tests sample set and check sample set carries out stratified sampling and coordinates to sample.
One, stratified sampling：
The number of plies of stratified sampling needs to refer to estimation accuracy of measurement, cost of sampling etc., can be selected in 3 to 10 layers.And in view of taking out
Sample cost, practical sampling difficulty and experiment will investigate target, choose candidate dimension as sublayer attribute, or can also basis
Experience and information gain analysis select the higher sublayer attribute of the degree of correlation, specific sublayer attribute can according to actual conditions and
It is fixed.
For example, 4 sublayer attributes can be taken, first sublayer attribute segments number is A, and second sublayer attribute segments number is
G, third sublayer attribute segments number are L, and the 4th sublayer attribute segments number is H, and it is M, total size to enable sample drawn sum
For N, then final layering sum is A × G × L × H.
Since each level is at least there are two representative sample, so M ＞ max { A, G, L, H }, and M ＞ A × G × L × H, because
This, samples using by sublayer ratio, and the sample number that any sublayer extracts is：
Wherein, 1,2,3 or 4 i, j 1,2,3 or 4, k 1,2,3 or 4, t 1,2,3 or 4, AiGjLkHtRefer to A × G × L
A sublayer in × H sublayer,For the sum of the sublayer.
In addition, stratified sampling can also use other algorithms other than above-mentioned algorithm, such as interior graceful distribution.
In concrete scene, the experiment sample set of 10,000 samples and the check sample set of 1,000,000 samples have user year
Three sublayer attributes of number in age, user's gender and read access game application weekly, age of user are divided into 0-18 Sui, 19-35
Year, 36 to 50 years old three phases, user's gender is divided into man, two kinds of female, then in experiment sample set and check sample set
When the number of the game application of read access weekly of user is sampled comparison, it can be carried out according to age of user, user's gender
Layering, in any sublayer, the age of user of experiment sample and check sample, user's gender are consistent, in this base
On plinth, the number of read access game application weekly is compared, comparativity can be had more.
Be layered obtain be 0-18 Sui man, 19-35 Sui man, 36 to 50 years old men, 0-18 Sui female, 19-35 Sui female, 36 to 50 years old
Six sublayers of female, distribution are as shown in table 1 below.Wherein, understand for convenience, the layering amount in this scene is all made of integer progress
Example.
Table 1
| Sublayer | Experiment sample layering amount (a) | Check sample layering amount (a) |
| 0-18 Sui man | 1400 | 150000 |
| 19-35 Sui man | 3500 | 300000 |
| 36 to 50 years old men | 1600 | 150000 |
| 0-18 Sui female | 800 | 80000 |
| 19-35 Sui female | 2000 | 250000 |
| 36 to 50 years old female | 700 | 70000 |
Two, coordinate sampling：
Consistency check further is done to experiment sample set and check sample set, is then receiving certain error
Under the premise of extract and be distributed the balanced sample collection consistent with quantity.
The sample total N=∑s N of experiment sample seto, o=1,2 ..., A × G × L × H, order expectation check sample total amount
For EM, then according to each layer sampling information as shown in table 2 below, for all sublayers, if EMo<No<Mo, then each layer can be with
By amount of sampling EMoIt extracts.If being unsatisfactory for EMo<No<Mo, then can choose sampling bottleneck and be coordinated.
Table 2
There is the layering amount of several layerings especially small in check sample set, layering amount it is expected in the respectively less than control of itself,
It can then select the layering of layering amount minimum as sampling bottleneck wherein, the expectation layering amount of each layer is carried out by sampling bottleneck
It reduces in proportion.Wherein, layering amount and the ratio for compareing desired layering amount are small, and layering amount is minimum.
Reduction can be made inferences to sampling bottleneck：
User's expectation sampling ratio be EM/N, final sampling ratio then byIt determines, ordinary circumstance should meetCoordinating the stratified sample amount after sampling is
In addition, will appear the overall certain layering amounts of control on rare occasions is 0, if taking such data as bottle
Neck, whole zoom ratio are 0, cannot take out any sample.Therefore it can find minimum more than 0Ratio as sampling
Bottleneck.
Or, it is also possible to it will appear the case where overall certain layering amounts of control are close to 0, if such data is taken to make
For bottleneck, for whole zoom ratio close to 0, the sample of taking-up is few, influences practical comparison.At this point it is possible to take more than predetermined threshold
It is minimumRatio as sampling bottleneck, which is that pre-set can normally complete needed for balanced sampling
Minimum sampling ratio.
After balanced sampling, can carry out simple experiment statistics and Parameter analysis to balanced sample, such as analysis exposure and it is non-
The influence to conversion rate of products is exposed, it is of course also possible to the inspection of other statistical parameter consistency.
In conjunction with above-mentioned concrete scene, the sample number N of experiment sample set is 10,000, and the sample number M of check sample set is
1000000, then when judging whether to need to carry out it to coordinate sampling, it is necessary first to judge whether that layering amount it is expected in o layers of control
EMoBoth less than experiment sample layering amount No, and experiment sample layering amount NoLess than check sample layering amount Mo。
When given expectation check sample total amount EM is 4000, for 0-18 Sui male first sublayer, layering it is expected in control
Measure EM1Similarly 19-35 Sui man, 36 to 50 years old men, 0-18 Sui female, 19- is calculated in=4000 × 15000/100000=600
The control of 35 years old female, 36 to 50 years old female, five sublayers it is expected that layering amount is respectively 1200,600,320,1000 and 280, are satisfied by
Meet EMo<No<Mo, then can be directly according to EMoIt is sampled.
When given expectation check sample total amount EM is 10000, for 0-18 Sui man this first sublayer, EM1=10000
× 15000/100000=1500 similarly it is expected that layering amount calculates, due to min (M to the control of other sublayerso/EMo) it is 0.8,
Therefore the control of each sublayer can it is expected that layering amount reduces 0.8, obtain distribution results as shown in table 3 below.
Table 3
| Sublayer | Mo | EMo | No | Mo/EMo | Practical extraction amount Ro |
| 0-18 Sui man | 1400 | 1500 | 150000 | 0.93 | 1200 |
| 19-35 Sui man | 3500 | 3000 | 300000 | 1.17 | 2800 |
| 36 to 50 years old men | 1600 | 1500 | 150000 | 0.93 | 1200 |
| 0-18 Sui female | 800 | 800 | 80000 | 1 | 640 |
| 19-35 Sui female | 2000 | 2500 | 250000 | 0.8 | 2000 |
| 36 to 50 years old female | 700 | 700 | 70000 | 1 | 560 |
The present invention is by introducing stratified sampling, the overall and control inconsistent distribution of entity is tested in processing, is allowing certain miss
Total amount is extracted in the case of difference and is distributed consistent balanced sample, compared to experimental effect analysis.To test in practice
Comparison provides good theoretical foundation and method reference, ensure that the objectivity and science of Experimental comparison.
Fig. 4 shows the block diagram for the sampling apparatus that one embodiment of the present of invention provides.
As shown in figure 4, the sampling apparatus 400 that one embodiment of the present of invention provides, including：Stratified sampling amount acquiring unit
402, first selecting unit 404, coordination amount of sampling acquiring unit 406 and sample process unit 408.
Wherein, stratified sampling amount acquiring unit 402 is used for according to specified sublayer attribute, respectively to experiment sample set and
Check sample set carries out layered shaping, to obtain the first amount of sampling and check sample of each sublayer of experiment sample set
Second amount of sampling of each sublayer of set.
First selecting unit 404 is used to be not more than the first amount of sampling when the second amount of sampling of each sublayer, alternatively, first takes out
When sample amount it is expected layering amount no more than control, selected in the proportionate relationship from the first amount of sampling of each sublayer with the second amount of sampling
Meet the first object proportionate relationship of the first specified requirements.
Coordination amount of sampling acquiring unit 406 is used for the first amount of sampling according to first object proportionate relationship and each sublayer,
Obtain the third amount of sampling of each sublayer.
Sample process unit 408 is for being sampled experiment sample set and check sample set according to third amount of sampling
Processing.
In the above embodiment of the present invention, optionally, sample process unit 408 is additionally operable to：Second when each sublayer takes out
Sample amount be more than the first amount of sampling, and the first amount of sampling be more than control it is expected layering amount when, based on control it is expected layering amount to experiment
Sample set and check sample set are sampled processing.
In the above embodiment of the present invention, optionally, sampling apparatus 400 further includes：Judging unit 410, in first choice
After the selection of unit 404 meets the first object proportionate relationship of the first specified requirements, judge whether first object proportionate relationship is full
The second specified requirements of foot；Coordinate amount of sampling acquiring unit 406 to be specifically used for：If first object proportionate relationship meets the second specified item
Part executes first amount of sampling of the step according to first object proportionate relationship and each sublayer, obtains the third sampling of each sublayer
Amount, and if first object proportionate relationship be unsatisfactory for the second specified requirements, from other in addition to first object proportionate relationship
Selection meets the second target proportion relationship of the first specified requirements in proportionate relationship, according to the second target proportion relationship and per height
First amount of sampling of layer, obtains the third amount of sampling of each sublayer.
In the above embodiment of the present invention, optionally, the first specified requirements is the ratio of the second amount of sampling and the first amount of sampling
Value is minimum；Or the ratio maximum that first specified requirements is the first amount of sampling and the second amount of sampling.
In the above embodiment of the present invention, optionally, second specified requirements is the first object proportionate relationship and 0
Difference be more than specified threshold；Or second specified requirements is equal to finger for the difference of the first object proportionate relationship and 0
Determine threshold value.
In the above embodiment of the present invention, optionally, stratified sampling amount acquiring unit 402 is specifically used for：From experiment sample
In set extract with sublayer attribute sample, the total amount of the sample as the first amount of sampling, and, from check sample set
The sample with sublayer attribute is extracted, the total amount of the sample is as the second amount of sampling.
Fig. 5 shows the block diagram for the terminal that one embodiment of the present of invention provides.
As shown in figure 5, terminal 500 according to an embodiment of the invention, including the sampling apparatus 400 shown in Fig. 4, because
This, the terminal 500 have and Fig. 4 shown in 400 identical technique effect of sampling apparatus, details are not described herein.
Fig. 6 shows the block diagram for the terminal that an alternative embodiment of the invention provides.
As shown in fig. 6, terminal 600 may include the processor 602 being connect with one or more data storage facilities, the data
Storage tool may include storage medium 604 and internal storage location 606.Terminal 600 can also include that input interface 608 and output connect
Mouth 610, for being communicated with another device or system.It is storable in by the CPU6022 of processor 602 program codes executed
In storage medium 604 or internal storage location 606.
Processor 602 in terminal 600 calls the program code for being stored in storage medium 604 or internal storage location 606, can hold
The following each step of row：
According to specified sublayer attribute, layered shaping is carried out to experiment sample set and check sample set respectively, to obtain
Obtain the first amount of sampling of each sublayer of experiment sample set and the second amount of sampling of each sublayer of check sample set；
When each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling no more than control it is expected
When layering amount, meet the first specified requirements from selection in the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
First object proportionate relationship；
According to the first amount of sampling of the first object proportionate relationship and each sublayer, the third sampling of each sublayer is obtained
Amount；
Processing is sampled to the experiment sample set and the check sample set according to third amount of sampling.
In a concrete implementation scheme, processor 602 can also carry out：
When each sublayer the second amount of sampling be more than the first amount of sampling, and the first amount of sampling be more than control it is expected layering amount
When, it is expected that layering amount is sampled processing to the experiment sample set and the check sample set based on control.
In a concrete implementation scheme, processor 602 can also carry out：
Meet the first specified requirements from selection in the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
After first object proportionate relationship, judge whether the first object proportionate relationship meets the second specified requirements；
If the first object proportionate relationship meets the second specified requirements, executes step and closed according to the first object ratio
First amount of sampling of system and each sublayer, obtains the third amount of sampling of each sublayer；
If the first object proportionate relationship is unsatisfactory for the second specified requirements, from except the first object proportionate relationship with
Selection meets the second target proportion relationship of the first specified requirements in other outer proportionate relationships, according to second target proportion
First amount of sampling of relationship and each sublayer obtains the third amount of sampling of each sublayer.
Wherein, first specified requirements is the ratio minimum of the second amount of sampling and the first amount of sampling；Or described first
Specified requirements is the ratio maximum of the first amount of sampling and the second amount of sampling.Second specified requirements is the first object ratio
Relationship with 0 difference be more than specified threshold；Or second specified requirements is the first object proportionate relationship and 0 difference
Equal to specified threshold.
In a concrete implementation scheme, processor 602 can be used for executing：
Extract the sample with the sublayer attribute from the experiment sample set, the total amount of the sample is as described the
One amount of sampling；And
Extract the sample with the sublayer attribute from the check sample set, the total amount of the sample is as described the
Two amount of sampling.
Technical scheme of the present invention is described in detail above in association with attached drawing, technical solution through the invention is capable of providing
A kind of new sample mode, the sample drawn distribution situation balanced sample consistent with sample total, such balanced sample have more
There is comparativity, compared and investigated using such balanced sample, the reliability of sampling comparison result can be further increased, carried
Rise the objectivity and science of Experimental comparison.
It will be appreciated that though may describe to sample using term first, second, third, etc. in embodiments of the present invention
Amount, but these amount of sampling should not necessarily be limited by these terms.These terms are only used for amount of sampling being distinguished from each other out.For example, not taking off
In the case of from range of embodiment of the invention, the first amount of sampling can also be referred to as the second amount of sampling, similarly, the second amount of sampling
It can also be referred to as the first amount of sampling.
Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination " or " in response to detection ".Similarly, depend on context, phrase " if it is determined that " or " if detection
(condition or event of statement) " can be construed to " when determining " or " in response to determination " or " when the detection (condition of statement
Or event) when " or " in response to detection (condition or event of statement) ".
It should be noted that terminal involved in the embodiment of the present invention can include but is not limited to personal computer
(Personal Computer, PC), personal digital assistant (Personal Digital Assistant, PDA), wireless handheld
Equipment, tablet computer (Tablet Computer), mobile phone, MP3 player, MP4 players etc..
In several embodiments provided by the present invention, it should be understood that disclosed systems, devices and methods, it can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of division of logic function, formula that in actual implementation, there may be another division manner, for example, multiple units or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be by some interfaces, device or unit it is indirect
Coupling or communication connection can be electrical, machinery or other forms.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also
It is that each unit physically exists alone, it can also be during two or more units be integrated in one unit.Above-mentioned integrated list
The form that hardware had both may be used in member is realized, can also be realized in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can be stored in one and computer-readable deposit
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that device (can be personal computer, server or network equipment etc.) or processor (Processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes：USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or CD etc. it is various
The medium of program code can be stored.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
With within principle, any modification, equivalent substitution, improvement and etc. done should be included within the scope of protection of the invention god.
Claims (12)
1. a kind of methods of sampling, which is characterized in that including：
According to specified sublayer attribute, layered shaping is carried out to experiment sample set and check sample set respectively, to obtain reality
Test the first amount of sampling of each sublayer of sample set and the second amount of sampling of each sublayer of check sample set；
When each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling no more than control it is expected layering
When amount, meet the first of the first specified requirements from selection in the proportionate relationship of the first amount of sampling of each sublayer and the second amount of sampling
Target proportion relationship；
According to the first amount of sampling of the first object proportionate relationship and each sublayer, the third amount of sampling of each sublayer is obtained；
Processing is sampled to the experiment sample set and the check sample set according to third amount of sampling.
2. according to the method described in claim 1, it is characterized in that, further including：
When the second amount of sampling of each sublayer is more than the first amount of sampling, and the first amount of sampling is more than control expectation layering amount, base
It is expected that layering amount is sampled processing to the experiment sample set and the check sample set in control.
3. according to the method described in claim 1, it is characterized in that, from the first amount of sampling of each sublayer and the second amount of sampling
After selection meets the first object proportionate relationship of the first specified requirements in proportionate relationship, further include：
Judge whether the first object proportionate relationship meets the second specified requirements；
If the first object proportionate relationship meets the second specified requirements, execute step according to the first object proportionate relationship and
First amount of sampling of each sublayer, obtains the third amount of sampling of each sublayer；
If the first object proportionate relationship is unsatisfactory for the second specified requirements, from addition to the first object proportionate relationship
Selection meets the second target proportion relationship of the first specified requirements in other proportionate relationships, according to the second target proportion relationship
With the first amount of sampling of each sublayer, the third amount of sampling of each sublayer is obtained.
4. according to the method in any one of claims 1 to 3, which is characterized in that first specified requirements is the second pumping
Sample amount and the ratio of the first amount of sampling are minimum；Or
First specified requirements is the ratio maximum of the first amount of sampling and the second amount of sampling.
5. according to the method described in claim 3, it is characterized in that, second specified requirements closes for the first object ratio
It is and 0 difference is more than specified threshold；Or
Second specified requirements is equal to specified threshold for the difference of the first object proportionate relationship and 0.
6. according to the method described in claim 1, it is characterized in that, according to specified sublayer attribute, respectively to experiment sample collection
It closes and check sample set carries out layered shaping, including：
The sample with the sublayer attribute is extracted from the experiment sample set, the total amount of the sample is taken out as described first
Sample amount；And
The sample with the sublayer attribute is extracted from the check sample set, the total amount of the sample is taken out as described second
Sample amount.
7. a kind of sampling apparatus, which is characterized in that including：
Stratified sampling amount acquiring unit, according to specified sublayer attribute, respectively to experiment sample set and check sample set into
Row layered shaping, to obtain the first amount of sampling of each sublayer of experiment sample set and each sublayer of check sample set
The second amount of sampling；
First selecting unit, when each sublayer the second amount of sampling be not more than the first amount of sampling, alternatively, the first amount of sampling is not more than
When layering amount it is expected in control, in the proportionate relationship from the first amount of sampling of each sublayer with the second amount of sampling selection meet the first finger
The first object proportionate relationship of fixed condition；
Coordinate amount of sampling acquiring unit, according to the first amount of sampling of the first object proportionate relationship and each sublayer, obtains every
The third amount of sampling of a sublayer；
Sample process unit is sampled place according to third amount of sampling to the experiment sample set and the check sample set
Reason.
8. device according to claim 7, which is characterized in that the sample process unit is additionally operable to：
When the second amount of sampling of each sublayer is more than the first amount of sampling, and the first amount of sampling is more than control expectation layering amount, base
It is expected that layering amount is sampled processing to the experiment sample set and the check sample set in control.
9. device according to claim 7, which is characterized in that further include：
Judging unit selects after meeting the first object proportionate relationship of the first specified requirements in the first selecting unit, sentences
Whether the first object proportionate relationship of breaking meets the second specified requirements；
The coordination amount of sampling acquiring unit is specifically used for：
If the first object proportionate relationship meets the second specified requirements, execute step according to the first object proportionate relationship and
First amount of sampling of each sublayer, obtains the third amount of sampling of each sublayer, and if the first object proportionate relationship it is discontented
The second specified requirements of foot, then selection satisfaction first is specified from other proportionate relationships in addition to the first object proportionate relationship
Second target proportion relationship of condition obtains every according to the first amount of sampling of the second target proportion relationship and each sublayer
The third amount of sampling of a sublayer.
10. the device according to any one of claim 7 to 9, which is characterized in that first specified requirements is the second pumping
Sample amount and the ratio of the first amount of sampling are minimum；Or
First specified requirements is the ratio maximum of the first amount of sampling and the second amount of sampling.
11. device according to claim 9, which is characterized in that second specified requirements is the first object ratio
Relationship with 0 difference be more than specified threshold；Or
Second specified requirements is equal to specified threshold for the difference of the first object proportionate relationship and 0.
12. device according to claim 7, which is characterized in that the stratified sampling amount acquiring unit is specifically used for：
The sample with the sublayer attribute is extracted from the experiment sample set, the total amount of the sample is taken out as described first
Sample amount, and, the sample with the sublayer attribute is extracted from the check sample set, described in the total amount of the sample is used as
Second amount of sampling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710035012.1A CN108319611B (en) | 2017-01-17 | 2017-01-17 | Sampling method and sampling device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710035012.1A CN108319611B (en) | 2017-01-17 | 2017-01-17 | Sampling method and sampling device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108319611A true CN108319611A (en) | 2018-07-24 |
| CN108319611B CN108319611B (en) | 2022-03-11 |
Family
ID=62892226
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710035012.1A Active CN108319611B (en) | 2017-01-17 | 2017-01-17 | Sampling method and sampling device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108319611B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110825783A (en) * | 2019-10-31 | 2020-02-21 | 深圳前海微众银行股份有限公司 | Data sampling method, device, equipment and storage medium |
| CN111865753A (en) * | 2019-04-26 | 2020-10-30 | 腾讯科技（深圳）有限公司 | Method and device for determining parameters of media information, storage medium and electronic device |
| CN112579983A (en) * | 2021-03-01 | 2021-03-30 | 深圳市城市交通规划设计研究中心股份有限公司 | Travel survey sampling method and device |
| CN112711643A (en) * | 2019-10-25 | 2021-04-27 | 北京达佳互联信息技术有限公司 | Training sample set obtaining method and device, electronic equipment and storage medium |
| CN113312554A (en) * | 2021-06-15 | 2021-08-27 | 北京百度网讯科技有限公司 | Method and device for evaluating recommendation system, electronic equipment and medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040049492A1 (en) * | 2002-09-09 | 2004-03-11 | Lucent Technologies Inc. | Distinct sampling system and a method of distinct sampling for a database |
| CN104965976A (en) * | 2015-06-12 | 2015-10-07 | 北京京东尚科信息技术有限公司 | Sampling method and device |
-
2017
- 2017-01-17 CN CN201710035012.1A patent/CN108319611B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040049492A1 (en) * | 2002-09-09 | 2004-03-11 | Lucent Technologies Inc. | Distinct sampling system and a method of distinct sampling for a database |
| CN104965976A (en) * | 2015-06-12 | 2015-10-07 | 北京京东尚科信息技术有限公司 | Sampling method and device |
Non-Patent Citations (3)
| Title |
|---|
| ROBERTO CID FERNANDES: "A probabilistic formulation for empirical population synthesis: sampling", 《MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY》 * |
| 张峰等: "基于分层抽样的高速网络吞吐率测量", 《吉林大学学报(信息科学版)》 * |
| 袁兴伟等: "分层抽样误差分析及其在渔业统计中的应用", 《海洋渔业》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111865753A (en) * | 2019-04-26 | 2020-10-30 | 腾讯科技（深圳）有限公司 | Method and device for determining parameters of media information, storage medium and electronic device |
| CN112711643A (en) * | 2019-10-25 | 2021-04-27 | 北京达佳互联信息技术有限公司 | Training sample set obtaining method and device, electronic equipment and storage medium |
| CN112711643B (en) * | 2019-10-25 | 2023-10-10 | 北京达佳互联信息技术有限公司 | Training sample set acquisition method and device, electronic equipment and storage medium |
| CN110825783A (en) * | 2019-10-31 | 2020-02-21 | 深圳前海微众银行股份有限公司 | Data sampling method, device, equipment and storage medium |
| CN112579983A (en) * | 2021-03-01 | 2021-03-30 | 深圳市城市交通规划设计研究中心股份有限公司 | Travel survey sampling method and device |
| CN112579983B (en) * | 2021-03-01 | 2021-07-30 | 深圳市城市交通规划设计研究中心股份有限公司 | Travel survey sampling method and device |
| CN113312554A (en) * | 2021-06-15 | 2021-08-27 | 北京百度网讯科技有限公司 | Method and device for evaluating recommendation system, electronic equipment and medium |
| CN113312554B (en) * | 2021-06-15 | 2023-11-03 | 北京百度网讯科技有限公司 | Method and device for evaluating recommendation system, electronic equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108319611B (en) | 2022-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108319611A (en) | The methods of sampling and sampling apparatus | |
| CN111339436B (en) | Data identification method, device, equipment and readable storage medium | |
| EP4198775A1 (en) | Abnormal user auditing method and apparatus, electronic device, and storage medium | |
| CN104199898B (en) | Acquisition methods and device, the method for pushing and device of a kind of attribute information | |
| CN106874253A (en) | Recognize the method and device of sensitive information | |
| Mauguen et al. | Using the Lorenz curve to characterize risk predictiveness and etiologic heterogeneity | |
| TW201710993A (en) | Method, apparatus and system for detecting fraudulent software promotion | |
| CN106776925B (en) | Method, server and system for predicting gender of mobile terminal user | |
| Allignol et al. | Estimating summary functionals in multistate models with an application to hospital infection data | |
| CN110362601A (en) | Mapping method, device, equipment and the storage medium of metadata standard | |
| CN111611390B (en) | Data processing method and device | |
| CN109753275B (en) | Recommendation method and device for application programming interface, storage medium and electronic equipment | |
| CN108196273A (en) | A kind of position location satellite selection method of integrated navigation system | |
| CN108512883A (en) | A kind of information-pushing method, device and readable medium | |
| CN107766467A (en) | Information detection method and device, electronic equipment and storage medium | |
| CN115690672A (en) | Abnormal image recognition method and device, computer equipment and storage medium | |
| CN107767155B (en) | Method and system for evaluating user portrait data | |
| CN108932658B (en) | Data processing method, device and computer readable storage medium | |
| CN109657840A (en) | Decision tree generation method, device, computer readable storage medium and electronic equipment | |
| CN112966756A (en) | Visual access rule generation method and device, machine readable medium and equipment | |
| CN110210884B (en) | Method, device, computer equipment and storage medium for determining user characteristic data | |
| CN113891323B (en) | WiFi-based user tag acquisition system | |
| CN110163659A (en) | Data estimation method, apparatus, equipment and storage medium | |
| CN111209925A (en) | Gender prediction method, device and computer-readable storage medium | |
| Sasaki et al. | Estimating sexual size dimorphism in fossil species from posterior probability densities |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |