CN104102813A - White noise compensation method and device - Google Patents
White noise compensation method and device Download PDFInfo
- Publication number
- CN104102813A CN104102813A CN201410245026.2A CN201410245026A CN104102813A CN 104102813 A CN104102813 A CN 104102813A CN 201410245026 A CN201410245026 A CN 201410245026A CN 104102813 A CN104102813 A CN 104102813A
- Authority
- CN
- China
- Prior art keywords
- sample
- attribute vector
- value
- described attribute
- measured value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Complex Calculations (AREA)
Abstract
The invention relates to a white noise compensation method and a white noise compensation device. The white noise compensation method is characterized by comprising the following steps: calculating the average value of measurement values of an attribute vector of each sample in sample space, and taking the average value as an estimation value of the attribute vector; calculating the average value of the estimation values of the attribute vectors of all samples; complementing the number of the measurement values of the attribute vector of the sample with the number of the measurement values of the attribute vector smaller than a preset number to the preset number by taking the average value of the estimation values of the attribute vectors of all samples as the measurement value of the attribute vector of the sample; recalculating the average value of the measurement values of the attribute vector of the sample of which the number of the measurement values of the attribute vectors is complemented, and taking the recalculated average value as the estimation value.
Description
Technical field
The present invention relates to field of information processing, particularly the method for the statistical treatment in field of information processing and system.
Background technology
Current, in various fields, all relate to the statistics of sample attribute value, such as the statistics of the sample attribute value all relating in fields such as internet industrys.
But when the measured value of sample attribute value is less, statistics cannot reflect actual conditions sometimes truly, even there will be statistics allow the elusive situation of people.Particularly in recommendation of personalized information field, behavioral data to user arranges, when generating the interactive relation matrix between user and article according to user's operation behavior, conventionally exist certain user's operation behavior very few, the unsettled situation of statistics.
Summary of the invention
The present invention completes in view of the above problems, its object of the present invention is to provide in the situation that work as the less method of utilizing white noise to compensate of measured value of sample attribute value, realization, in the situation that the measured value of sample attribute value is less, is obtained stable sample attribute Data-Statistics result.
According to a kind of white noise compensation method of the present invention, it is characterized in that, comprise the following steps: the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; Calculate the mean value of estimated value of the described attribute vector of all samples; For the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of estimated value of described attribute vector of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
According to a kind of white noise compensation method of the present invention, it is characterized in that, comprise the following steps: the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; Calculate the mean value of measured value of all described attribute vectors of all samples; For the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of measured value of all described attribute vectors of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
According to a kind of white noise compensation system of the present invention, it is characterized in that, comprising: the estimated value computing unit of each sample attribute vector, the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; The average calculation unit of sample attribute vector, calculates the mean value of estimated value of the described attribute vector of all samples; White noise is supplied unit, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of estimated value of described attribute vector of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Supply the estimated value computing unit of white noise sample attribute vector, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
According to a kind of white noise compensation system of the present invention, it is characterized in that, comprising: the estimated value computing unit of each sample attribute vector, the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; The average calculation unit of all sample attribute vectors, calculates the mean value of measured value of all described attribute vectors of all samples; White noise is supplied unit, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of measured value of all described attribute vectors of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Supply white noise sample attribute vector estimated value computing unit, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
According to white noise compensation method of the present invention and device, even if also can access good statistics effect in the situation that of sample attribute vector measurement value negligible amounts, can be applicable to any field of information processing that need to carry out sample attribute Data-Statistics, efficiently solve the problem of stablizing statistics in statistics.
Accompanying drawing explanation
Fig. 1 illustrates the process flow diagram of the white noise compensation method of embodiment 1;
Fig. 2 illustrates the process flow diagram of another white noise compensation method of embodiment 1;
Fig. 3 illustrates the process flow diagram of the white noise compensation method of embodiment 2;
Fig. 4 illustrates the process flow diagram of another white noise compensation method of embodiment 2;
Fig. 5 illustrates the process flow diagram of the white noise compensation method of embodiment 3;
Fig. 6 illustrates the process flow diagram of another white noise compensation method of embodiment 3;
Fig. 7 is the block diagram that white noise compensation system is shown;
Fig. 8 is the block diagram that another white noise compensation system is shown.
Embodiment
The sample group that the present invention be directed to the sample attribute vector measurement value negligible amounts of sample space is obtained the attribute vector estimated value of sample comparatively exactly.
First, calculate the estimated value of each sample attribute vector.Utilize the attribute vector measured value of known sample, ask for the average of attribute vector measured value as the estimated value of the attribute vector of each sample.
Then, according to the estimated value of each sample attribute vector, obtain the average of the estimated value of sample attribute vector.
Then, for attribute vector number of measurement values in sample, be less than the sample of some, utilize the average of the estimated value of resulting attribute vector to supply sample attribute vector measurement value number as sample attribute vector measurement value.
Finally, for the sample of supplying attribute vector number of measurements, recalculate the estimated value of new sample attribute vector.
Be specifically described for example below.
Embodiment 1
Factory adds up produce three batches of part lifes, but because sampling sample size is different, first batch more with second batch sub-sampling sample, the 3rd batch of sampling sample is less, in order further to understand the situation of the 3rd batch of sampling sample, carry out white noise compensation, the 3rd batch of sampling sample complemented to the quantity M that can substantially reflect accurate estimated value.Concrete operations illustrate as follows with reference to figure 1.
First, as shown in the step S11 of Fig. 1, calculate respectively the mean lifetime t1 of the part of first production, the mean lifetime t2 of part of second batch time production and the mean lifetime t3 of the part of the 3rd batch of production.
Then, as shown in step S12, calculate the mean lifetime t of these three batches all parts of producing.
Then,, as shown in step S13, with the sample number that the mean lifetime t of all parts removes to supply the 3rd series-produced part, making the 3rd batch of number of samples of producing is M.
Finally, as shown in step S14, recalculate the mean lifetime t1 ' of the 3rd series-produced part, as the mean lifetime of the 3rd series-produced part.
Here, as shown in Figure 2, replace step S12, also can not calculate the mean lifetime t of part used, but as shown in step S22, calculate the mean value of mean lifetime of the part of three batches, i.e. the value of (t1+t2+t3)/3 gained, and with this mean value, as part life, go the sample number of supplying the 3rd series-produced part to M in step S23.
Embodiment 2
With reference to figure 3, take the quality of evaluating and testing a film by the scoring of the film on web film describes as example.As shown in table 1, for example there are three films, be respectively film A, film B, film C, the scoring of film A is had to 2 people, the scoring of film B is had to 5 people, the scoring of film C is had to 9 people, mark is 0 minute to 10 minutes, 10 are divided into full marks.First row represents film, and spectators' scoring is shown in 2-10 list.
Table 1
| ? | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Film A | 5 | 7 | ? | ? | ? | ? | ? | ? | ? |
| Film B | 6 | 8 | 7 | 6 | 5 | ? | ? | ? | ? |
| Film C | 8 | 7 | 8 | 6 | 5 | 7 | 8 | 6 | 8 |
First, as shown in the step S31 of Fig. 3, calculate the average mark of every film, the average mark of film A divides for (5+7)/2=6, the average mark of film B divides for (6+8+7+6+5)/5=6.4, and the average mark of film C divides for (8+7+8+6+5+7+8+6+8)/9=6.
Then,, as shown in step S32, calculate the average mark (5+7+6+8+7+6+5+8+7+8+6+5+7+8+6+8)/(2+5+9)=6.6875 of all film marks.
As shown in step S33, at film A and the B lower than 6 people to film pricer number, utilize the average mark 6.6875 of all films to supply sample to 6, obtain following table 2.
Table 2
| ? | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Film A | 5 | 7 | 6.6875 | 6.6875 | 6.6875 | 6.6875 | ? | ? | ? |
| Film B | 6 | 8 | 7 | 6 | 5 | 6.6875 | ? | ? | ? |
| Film C | 8 | 7 | 8 | 6 | 5 | 7 | 8 | 6 | 8 |
As shown in step S34, reappraise the new score value of every film, the average mark of film A divides for (5+7+6.6875+6.6875+6.6875+6.6875)/6=6.458, the average mark of film B divides for (6+8+7+6+5+6.6875)/6=6.45, and the average mark of film C divides for (8+7+8+6+5+7+8+6+8)/9=6.Therefore to the fractional value of the scoring of film A, being 6.458, is 6.45 to the fractional value of the scoring of film B, to the fractional value of the scoring of film C, is 6.
In above-mentioned, as shown in Figure 4, the step S32 that also can replace Fig. 3 calculates the step of all film average marks, but as shown in step S42, calculate the average mark to all film average marks, i.e. (6+6.4+6)/3=6.13, and in step S43 by the 6.13 evaluation numbers of removing to supply film A and film B as fractional value.
By the above-mentioned sample for the discontented sufficient predetermined number of numerical value number in each sample, by supplying sample size to a certain predetermined value, thereby obtain stablizing the sample estimated value of statistics.
Embodiment described above has carried out sample to supply in the situation that there is an attribute.Also can be that attribute number does not meet predetermined number when carrying out sample statistics with a plurality of independent attributes sample carries out sample and supplies.
Embodiment 3
With reference to 5 pairs of situations that exist a plurality of independent attributes to carry out sample statistics of figure, describe.In fictitia, on some websites, have a lot of films, marking with watching duration is the attribute vector of film.Want now that the scoring and the user that determine certain film watch duration.Known have three users that this film is watched and marked now.Measured value about scoring on this website is respectively 7,5,8, about watching the measured value of duration to be respectively 1.4,1.6,1.5.Concrete condition is as shown in table 3.
Table 3
| ? | Scoring (full marks 10 minutes) | Watch duration (hour) |
| User 1 | 7 | 1.4 |
| User 2 | 5 | 1.6 |
| User 3 | 8 | 1.5 |
First as shown in step S51, calculate about the mean value of the scoring measured value of this film and the mean value of watching duration measured value, the mean value of measured value of marking is (7+5+8)/3=6.67, and the mean value of watching duration measured value is (1.4+1.6+1.5)/3=1.5.
The number of users that a known film is watched, more than 30, could reflect film in the actual scoring of this website and watch duration.But because above-mentioned this website only has 3 for the evaluation quantity of this film, evaluate quantity very few, therefore two of this film of this website attribute vectors are carried out to white noise compensation.
As shown in step S52, in order to predict that accurately scoring and the user of this film watches duration, this website is found on other similar websites, obtain each website to the scoring of this film and watch the mean value of duration, and calculate and comprise this website in the scoring of each interior website and watch the mean value of the mean value of duration measured value, as shown in table 4 below.
Table 4
| ? | Scoring (full marks 10 minutes) | Watch duration (hour) |
| Film | 6 | 1.2 |
As shown in step S53, utilize the mean value of above-mentioned mean value to carry out white noise compensation to two of this film of this website attributes, the quantity of attribute vector is complemented to 30.As shown in step S54, calculate the mean value of attribute vector measured value of this film of supplying after attribute vector measured value as estimated value, this website is predicted the scoring of this film and is watched duration to be:
([7，1.4]+[5，1.6]+[8，1.5]+27*[6，1.2])/30＝[6.0667，1.230]
So this film is predicted in this website, the scoring on this website is 6.0667 minutes, and watching duration is 1.230 hours.Thereby obtained prediction more accurately.
Also can replace the step S52 in Fig. 5, and shown in step S62 as shown in Figure 6, calculate the mean value of the measured value about attribute scoring of all websites, and calculate all websites about attribute, watch the mean value of the measured value of duration, and as shown in step S63, utilize the mean value of the above-mentioned measured value calculating to go to supply the number of attribute vector measured value of this film of this website, be 30, and utilize the attribute vector measured value of supplying after white noise to come the mean value of computation attribute vector measurement value as estimated value.
Fig. 7 illustrates a kind of white noise compensation system 700, comprising: the estimated value computing unit 701 of each sample attribute vector, and the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; The average calculation unit 702 of sample attribute vector, calculates the mean value of estimated value of the described attribute vector of all samples; White noise is supplied unit 703, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of estimated value of described attribute vector of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Supply the estimated value computing unit 704 of white noise sample attribute vector, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
Fig. 8 illustrates another kind of white noise compensation system 800, comprising: the estimated value computing unit 801 of each sample attribute vector, and the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector; The average calculation unit 802 of all sample attribute vectors, calculates the mean value of measured value of all described attribute vectors of all samples; White noise is supplied unit 803, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of measured value of all described attribute vectors of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number; Supply white noise sample attribute vector estimated value computing unit 804, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
According to above-mentioned, method and the device of white noise compensation has been described, it is conducive to stablize statistics, realizes and in the situation that number of samples is few, obtains good statistics.
The invention is not restricted to above embodiment, in the scope of its technical conceive, all belong to scope of the present invention.
Claims (4)
1. a white noise compensation method, is characterized in that, comprises the following steps:
The mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector;
Calculate the mean value of estimated value of the described attribute vector of all samples;
For the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of estimated value of described attribute vector of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number;
Recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
2. a white noise compensation method, is characterized in that, comprises the following steps:
The mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector;
Calculate the mean value of measured value of all described attribute vectors of all samples;
For the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of measured value of all described attribute vectors of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number;
Recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
3. a white noise compensation system, is characterized in that, comprising:
The estimated value computing unit of each sample attribute vector, the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector;
The average calculation unit of sample attribute vector, calculates the mean value of estimated value of the described attribute vector of all samples;
White noise is supplied unit, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of estimated value of described attribute vector of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number;
Supply the estimated value computing unit of white noise sample attribute vector, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
4. a white noise compensation system, is characterized in that, comprising:
The estimated value computing unit of each sample attribute vector, the mean value of the measured value of the attribute vector of each sample in calculating sample space is as the estimated value of described attribute vector;
The average calculation unit of all sample attribute vectors, calculates the mean value of measured value of all described attribute vectors of all samples;
White noise is supplied unit, for the number of measurements of described attribute vector, be less than the sample of predetermined number, utilize the mean value of measured value of all described attribute vectors of above-mentioned all samples as the measured value of the described attribute vector of sample, the number of measurements of the described attribute vector of sample is complemented to predetermined number;
Supply white noise sample attribute vector estimated value computing unit, recalculate the mean value of described attribute vector measured value of sample of the number of measurements of being supplied described attribute vector as estimated value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410245026.2A CN104102813A (en) | 2014-06-04 | 2014-06-04 | White noise compensation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410245026.2A CN104102813A (en) | 2014-06-04 | 2014-06-04 | White noise compensation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104102813A true CN104102813A (en) | 2014-10-15 |
Family
ID=51670958
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410245026.2A Pending CN104102813A (en) | 2014-06-04 | 2014-06-04 | White noise compensation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104102813A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107358946A (en) * | 2017-06-08 | 2017-11-17 | 南京邮电大学 | Speech-emotion recognition method based on section convolution |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080109428A1 (en) * | 2006-11-07 | 2008-05-08 | University Of Washington | Efficient top-k query evaluation on probabilistic data |
| CN101251850A (en) * | 2008-01-04 | 2008-08-27 | 杨虡 | Internet topics ranking system and method based on user prestige |
| US20090019009A1 (en) * | 2007-07-12 | 2009-01-15 | At&T Corp. | SYSTEMS, METHODS AND COMPUTER PROGRAM PRODUCTS FOR SEARCHING WITHIN MOVIES (SWiM) |
| CN103514304A (en) * | 2013-10-29 | 2014-01-15 | 海南大学 | Project recommendation method and device |
-
2014
- 2014-06-04 CN CN201410245026.2A patent/CN104102813A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080109428A1 (en) * | 2006-11-07 | 2008-05-08 | University Of Washington | Efficient top-k query evaluation on probabilistic data |
| US20090019009A1 (en) * | 2007-07-12 | 2009-01-15 | At&T Corp. | SYSTEMS, METHODS AND COMPUTER PROGRAM PRODUCTS FOR SEARCHING WITHIN MOVIES (SWiM) |
| CN101251850A (en) * | 2008-01-04 | 2008-08-27 | 杨虡 | Internet topics ranking system and method based on user prestige |
| CN103514304A (en) * | 2013-10-29 | 2014-01-15 | 海南大学 | Project recommendation method and device |
Non-Patent Citations (3)
| Title |
|---|
| "基于用户投票的排名算法（一）：Delicious和Hacker News";hzdjwl2010;《http://wenku.baidu.com/view/ffbe761452d380eb62946ddf.html?from=search》;20120424;第23-26页 * |
| HZDJWL2010: ""基于用户投票的排名算法（一）：Delicious和Hacker News"", 《HTTP://WENKU.BAIDU.COM/VIEW/FFBE761452D380EB62946DDF.HTML?FROM=SEARCH》 * |
| 张敏: ""对简单算术平均数与加权算术平均数的关系的质疑"", 《统计与决策》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107358946A (en) * | 2017-06-08 | 2017-11-17 | 南京邮电大学 | Speech-emotion recognition method based on section convolution |
| CN107358946B (en) * | 2017-06-08 | 2020-11-13 | 南京邮电大学 | Voice emotion recognition method based on slice convolution |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11887133B2 (en) | Methods and apparatus to generate electronic mobile measurement census data | |
| US11494636B2 (en) | Machine learning-based semiconductor manufacturing yield prediction system and method | |
| US9852440B2 (en) | Methods and systems for selecting content for display based on conversion probabilities of paths | |
| SG11201811395YA (en) | Method and device for controlling data risk | |
| KR102148248B1 (en) | Flagging of ad frames for automatic content recognition | |
| KR20220164621A (en) | Methods and apparatus to compensate impression data for misattribution and/or non-coverage by a database proprietor | |
| KR20150067224A (en) | Methods and apparatus to estimate demographics of users employing social media | |
| US20140279736A1 (en) | Method and system for mapping short term ranking optimization objective to long term engagement | |
| US10825030B2 (en) | Methods and apparatus to determine weights for panelists in large scale problems | |
| CN108322827B (en) | Method, system, and computer-readable storage medium for measuring a user's video preferences | |
| JP7085486B2 (en) | User behavior data processing method and equipment | |
| US20220012661A1 (en) | Methods, systems, articles of manufacture and apparatus to calibrate payload information | |
| US11711419B2 (en) | System and a method for multisession analysis | |
| US20240414075A1 (en) | System and a method for identifying client devices used by the same user | |
| CN104102813A (en) | White noise compensation method and device | |
| JP2011227720A (en) | Recommendation system, recommendation method and recommendation program | |
| CN110009161A (en) | Water supply forecast method and device | |
| CN111400512B (en) | Method and device for screening multimedia resources | |
| CN106375850B (en) | A kind of judgment method and device matching video | |
| US20170115332A1 (en) | Electricity consumption predicting system and electricity consumption predicting method applied for processing machine | |
| US11880643B2 (en) | Device and method for integrated circuit assistance design, and method for constructing electrical performance gradient model | |
| JPWO2016157927A1 (en) | Data processing apparatus, data processing method, and program | |
| JP2017117023A (en) | Data analysis device, data computing device, and program | |
| TW201620057A (en) | Yield rate determination method for wafer and method for multiple variable detection of wafer acceptance test | |
| CN113888199B (en) | Message ordering and publishing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20141015 |