CN107194505A - A kind of method and system that bus travel amount is predicted based on city big data - Google Patents
A kind of method and system that bus travel amount is predicted based on city big data Download PDFInfo
- Publication number
- CN107194505A CN107194505A CN201710324534.3A CN201710324534A CN107194505A CN 107194505 A CN107194505 A CN 107194505A CN 201710324534 A CN201710324534 A CN 201710324534A CN 107194505 A CN107194505 A CN 107194505A
- Authority
- CN
- China
- Prior art keywords
- bus
- mrow
- data
- travel amount
- forecast model
- 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
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 28
- 238000007477 logistic regression Methods 0.000 claims description 21
- 230000003068 static effect Effects 0.000 claims description 11
- 241001269238 Data Species 0.000 claims description 7
- 238000004140 cleaning Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims 1
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of method and system that bus travel amount is predicted based on city big data, the realization of wherein method includes:Sample data is extracted from the big data of city, according to the bus IC-card data on every public bus network, calculate the bus travel amount on every public bus network, and analysis is associated to sample data, feature set related to bus travel amount on every public bus network is extracted, bus travel amount and feature set is regard as input vector;Unsupervised pre-training is carried out to forecast model using input vector and has supervision to finely tune, the forecast model trained;The data of circuit to be predicted are gathered, the feature set of circuit to be predicted is extracted, the forecast model that the feature set input of circuit to be predicted is trained obtains the bus travel amount of circuit to be predicted.The present invention trains forecast model using input vector, the bus travel amount following available for circuit to be predicted is precisely predicted.
Description
Technical field
The invention belongs to smart city field, gone out more particularly, to one kind based on city big data prediction bus
The method and system of row amount.
Background technology
With quickly propelling for China's Development of China's Urbanization and increasing rapidly for economy, urban population and vehicle guaranteeding organic quantity swash
Increase, resident trip is increasingly frequent, the traffic problems of city faces severe.On the one hand, substantial amounts of urban transportation holding beyond road
Loading capability, causes traffic congestion, mass energy loss and environmental pollution.On the other hand, the low carrying capacity of Private Traffic instrument
The waste of ample resources is also result in relatively large occupancy path area.
Because urban public transport has large carrying capacity, conevying efficiency is high, the advantages of low and relative pollution of energy consumption is small, more
Start to actively push forward urban public transport construction come more areas, Public Transport Priority Development strategy is vigorously implemented, with Optimizing City
Transport structure, alleviate traffic congestion, economize on resources and reduce traffic carbon emission.The high efficiency operation of urban mass-transit system, not only takes
Certainly in the investment of large-scale basis facility, the management means of rational Operation Decision and science is relied more on.Wherein, efficiently utilize
One key factor of urban mass-transit system is exactly the travel amount of following certain time on real-time perception each public bus network.According to
Passenger's travel amount of following certain time on each public bus network, assists communications policy administrative department to formulate rational operator
Case, with Optimizing Urban Transportation, improves transportation dispatching level, meets the demand of people's trip.However, due to each area uniqueness
Geographical and cultural environment feature, the bus travel amount of city each department people is different because of when and where.Traditionally, traffic
Administrative department to each regional special investigation by counting the travel amount that each is regional, and these investigation are required for expending a large amount of
Time and expensive labour cost, and obtain that the cycle of data is long, passenger's trip situation of conversion zone is limited.
As can be seen here, passenger's travel amount of following certain time on each public bus network, consumption are predicted using conventional art
Take substantial amounts of time and expensive labour cost, and it is long to obtain the cycle of data.
The content of the invention
For the disadvantages described above or Improvement requirement of prior art, predicted the invention provides one kind based on city big data public
The method and system of automobile travel amount, its object is to extract input vector based on city big data, utilizes input vector pair altogether
Forecast model carries out unsupervised pre-training and has supervision to finely tune, and the forecast model trained utilizes the prediction mould trained
Type predicts bus travel amount, thus solves conventional art and predicts that the passenger of following certain time on each public bus network goes out
Row amount, takes a substantial amount of time and expensive labour cost, and obtains the cycle of data long technical problem.
To achieve the above object, predicted according to one aspect of the present invention there is provided one kind based on city big data public
The method of automobile travel amount, including:
(1) sample data is extracted from the big data of city, the sample data includes:Bus IC-card data, bus
Gps data, GPS data from taxi, city POI static datas and urban meteorological data;
(2) according to the bus IC-card data on every public bus network, the bus calculated on every public bus network goes out
Row amount, and analysis is associated to sample data, feature set related to bus travel amount on every public bus network is extracted,
It regard bus travel amount and feature set as input vector;
(3) unsupervised pre-training is carried out to forecast model using input vector and there is supervision to finely tune, what is trained is pre-
Survey model;
(4) the bus IC-card data of collection circuit to be predicted, bus GPS data, GPS data from taxi, city POI
Static data and urban meteorological data, extract the feature set of circuit to be predicted, and the feature set input of circuit to be predicted is trained
Forecast model, obtain the bus travel amount of circuit to be predicted.
Further, step (1) also includes cleaning redundancy in sample data and incomplete data.
Further, the specific implementation for being associated analysis in step (2) to sample data is:With on public bus network
An influence area is taken centered on each bus station, moving characteristic, geography in each influence area are analyzed using sample data
Incidence relation between feature and Meteorological Characteristics and resident trip, obtains feature set.
Further, step (3) includes:
(3-1) forecast model includes coding layer and logistic regression layer, and feature set and bus travel amount are forecast model
Input vector, bus trip premeasuring be forecast model output vector;In coding layer, nothing is carried out using input vector
Pre-training is supervised, the high-rise expression of feature set is obtained;
(3-2) is represented to obtain bus trip premeasuring, is utilized input in logistic regression layer, high-rise using feature set
The parameter of vector, bus trip premeasuring and forecast model sets up loss function, and coding layer is used for adjusting forecast model ginseng
Number minimizes loss function, and progress has supervision to finely tune, the forecast model trained.
Further, the specific implementation of step (3-1) is:Forecast model includes coding layer and logistic regression layer, special
Collection and bus travel amount be forecast model input vector, bus trip premeasuring for forecast model output to
Amount;Coding layer is made up of N layers of denoising autocoder, and the weight of every layer of input vector is minimized by reconstructing denoising autocoder
Structure error, and then unsupervised training is carried out to every layer of denoising autocoder, obtain the high-rise expression of feature set.
Further, coding layer is made up of 4 layers of denoising autocoder.
Further, the specific implementation of step (3-2) is:Logical layer is main to be calculated by a logistic regression supervised learning
Method is constituted, and high-rise by feature set represents input logic layer, obtains the bus trip premeasuring y on i-th line road(i), profit
With the input vector x on bus trip premeasuring and i-th public bus network(i), set up loss function:
Wherein, β is prediction model parameterses, D(s)The set of input vector is represented, P represents that is used for a predictablity rate
Function, Y represents the set of bus trip premeasuring;
Coding layer minimizes loss function for adjusting prediction model parameterses, and progress has supervision to finely tune, and is trained
Forecast model.
It is another aspect of this invention to provide that based on city big data predicting that bus travel amount is there is provided a kind of
System, including:
Sample data module is extracted, for extracting sample data from the big data of city, the sample data includes:Public transport
Car IC-card data, bus GPS data, GPS data from taxi, city POI static datas and urban meteorological data;
Input vector module is extracted, for according to the bus IC-card data on i-th public bus network, calculating i-th public affairs
Bus travel amount on intersection road, and sample data is associated on analysis, i-th public bus network of extraction and public transport vapour
The related feature set of car travel amount, regard bus travel amount and feature set as input vector;
Forecast model module is trained, for carrying out unsupervised pre-training to forecast model using input vector and having supervision micro-
Adjust, the forecast model trained;
Module is predicted, for gathering the bus IC-card data of circuit to be predicted, bus GPS data, taxi
Gps data, city POI static datas and urban meteorological data, extract the feature set of circuit to be predicted, by the spy of circuit to be predicted
The forecast model that collection input is trained, obtains the bus travel amount of circuit to be predicted.
Further, training forecast model module includes:
Unsupervised pre-training submodule, for carrying out unsupervised pre-training, forecast model includes coding layer and logistic regression
Layer, feature set and bus travel amount are the input vector of forecast model, and bus trip premeasuring is forecast model
Output vector;In coding layer, unsupervised pre-training is carried out using input vector, the high-rise expression of feature set is obtained;
There is supervision fine setting submodule, in logistic regression layer, high-rise using feature set to represent that obtaining bus goes out
Row premeasuring, loss function is set up using the parameter of input vector, bus trip premeasuring and forecast model, and coding layer is used
Loss function is minimized to adjust prediction model parameterses, progress has supervision to finely tune, the forecast model trained.
Have supervision fine setting submodule specific executive mode be:Logical layer is main by a logistic regression supervised learning algorithm
Constitute, high-rise by feature set represents input logic layer, obtain the bus trip premeasuring y on i-th line road(i), utilize
The input vector x that bus is gone on a journey on premeasuring and i-th public bus network(i), set up loss function:
Wherein, β is prediction model parameterses, D(s)The set of input vector is represented, P represents that is used for a predictablity rate
Function, Y represents the set of bus trip premeasuring;
Coding layer minimizes loss function for adjusting prediction model parameterses, and progress has supervision to finely tune, and is trained
Forecast model.
In general, by the contemplated above technical scheme of the present invention compared with prior art, it can obtain down and show
Beneficial effect:
(1) bus IC-card data, bus GPS data, GPS data from taxi, city POI are extracted from city static
Data and urban meteorological data equal samples data, help to analyse in depth the trip situation of city dweller;By analyzing sample number
Incidence relation between, helps to extract feature set closely related with bus travel amount on public bus network;Based on carrying
The feature set and bus travel amount taken, trains forecast model, helps precisely to predict the public vapour that circuit to be predicted is following
Car travel amount;
(2) preferred, cleaning sample data can effectively handle the invalid value and missing values in sample data, improve sample
Notebook data quality, and then improve training effectiveness and training accuracy rate;By to feature in each influence area on public bus network
Analysis, can obtain each distinctive information in section on public bus network, help to obtain more representational spy on public bus network
Collection;
(3) preferred, input vector high-rise represents have to sample data by what the unsupervised pre-training of coding layer was generated
More essential representativeness, and it is more conducive to the classification prediction of logistic regression layer;Have to forecast model after unsupervised pre-training
Supervision fine setting so that forecast model has more preferable estimated performance to bus travel amount;Coding layer to input vector by adding
Enter noise and reconstruct again, the ability to express of input vector can be significantly increased, obtain the high-rise expression of more robustness;Coding layer is used
The high-rise expression that 4 layers of denoising autocoder are abstract to input vector to be obtained, with stronger ability to express and is more conducive to carry
The performance of the classification prediction of high logistic regression layer;
(4) preferred, logistic regression layer is gone out by finely tuning the parameter of whole forecast model to minimize common root according to automobile altogether
The loss function that row premeasuring and input vector are built up, raising forecast model treats the pre- of the bus travel amount of prediction circuit
Survey performance.
Brief description of the drawings
Fig. 1 is a kind of flow chart for the method that bus travel amount is predicted based on city big data.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.As long as in addition, technical characteristic involved in each embodiment of invention described below
Not constituting conflict each other can just be mutually combined.
As shown in figure 1, a kind of method that bus travel amount is predicted based on city big data, including:
(1) sample data is extracted from the big data of city, the sample data includes:Bus IC-card data, bus
Gps data, GPS data from taxi, city POI static datas and urban meteorological data;
(2) according to the bus IC-card data on every public bus network, the bus calculated on every public bus network goes out
Row amount, and analysis is associated to sample data, feature set related to bus travel amount on every public bus network is extracted,
It regard bus travel amount and feature set as input vector;
(3) unsupervised pre-training is carried out to forecast model using input vector and there is supervision to finely tune, what is trained is pre-
Survey model;
(4) the bus IC-card data of collection circuit to be predicted, bus GPS data, GPS data from taxi, city POI
Static data and urban meteorological data, extract the feature set of circuit to be predicted, and the feature set input of circuit to be predicted is trained
Forecast model, obtain the bus travel amount of circuit to be predicted.
Further, step (1) also includes cleaning redundancy in sample data and incomplete data.
Further, due to the cross-domain very big region of a public bus network, and there is its unique feature in each region again,
Such as traffic characteristic, resident's moving characteristic, geographical and Meteorological Characteristics.In order to obtain the feature of each section of whole piece public bus network, step
(2) specific implementation for being associated analysis in sample data is:To be taken on public bus network centered on each bus station
One influence area, moving characteristic, geographical feature and Meteorological Characteristics and resident in each influence area are analyzed using sample data
Incidence relation between trip, obtains feature set.
It is preferred that, using randomly selecting 15 in all influence areas of the method for random sampling on every public bus network
Region, excavates the feature in this 15 influence areas to be combined into the feature set of whole piece circuit respectively.
Further, step (3) includes:
(3-1) forecast model includes coding layer and logistic regression layer, and feature set and bus travel amount are forecast model
Input vector, bus trip premeasuring be forecast model output vector;In coding layer, nothing is carried out using input vector
Pre-training is supervised, the high-rise expression of feature set is obtained;
(3-2) is represented to obtain bus trip premeasuring, is utilized input in logistic regression layer, high-rise using feature set
The parameter of vector, bus trip premeasuring and forecast model sets up loss function, and coding layer is used for adjusting forecast model ginseng
Number minimizes loss function, and progress has supervision to finely tune, the forecast model trained.
Further, the specific implementation of step (3-1) is:Forecast model includes input layer, coding layer, logistic regression
Layer and output layer, feature set and bus travel amount are the input vector of forecast model input layer, bus trip prediction
Measure as the output vector of forecast model output layer;Coding layer is made up of N layers of denoising autocoder, is compiled automatically by reconstructing denoising
Code device, obtains reconstruct vector sum reconstruct denoising autocoding parameter, minimizes the reconstructed error of every layer of input vector, and then to every
Layer denoising autocoder carries out unsupervised training, obtains the high-rise expression of feature set, denoising autocoding parameter θ and reconstruct are gone
Autocoding of making an uproar parameter θ ' be:
Wherein, x(i)For the input vector on bus trip premeasuring and i-th public bus network, x '(i)For to be public
The reconstruct that automobile is gone on a journey on premeasuring and i-th line road is vectorial, and n represents a shared n bar public bus networks, and L represents reconstructed error, L
(x(i), x '(i))=| | x '(i)-x(i)||2。
Further, coding layer is made up of 4 layers of denoising autocoder.
Further, the specific implementation of step (3-2) is:Logical layer is main to be calculated by a logistic regression supervised learning
Method is constituted, and high-rise by feature set represents input logic layer, obtains the bus trip premeasuring y on i-th line road(i), profit
With the input vector x on bus trip premeasuring and i-th public bus network(i), set up loss function:
Wherein, β is prediction model parameterses, D(s)The set of input vector is represented, P represents that is used for a predictablity rate
Function, Y represents the set of bus trip premeasuring;
Coding layer minimizes loss function for adjusting prediction model parameterses, and progress has supervision to finely tune, and is trained
Forecast model.
Further, prediction model parameterses include denoising autocoding parameter, reconstruct denoising autocoding parameter and reconstruct
The offset parameter of denoising autocoding function.
P (Y=y(i)|x(i), β) and P (Y=y can be designated as(i)|x(i), W, b),
Wherein, W represents denoising autocoding parameter and reconstruct denoising autocoding parameter, and b represents that reconstruct denoising is compiled automatically
The offset parameter of code function.
As it will be easily appreciated by one skilled in the art that the foregoing is merely illustrative of the preferred embodiments of the present invention, it is not used to
The limitation present invention, any modifications, equivalent substitutions and improvements made within the spirit and principles of the invention etc., it all should include
Within protection scope of the present invention.
Claims (10)
1. a kind of method that bus travel amount is predicted based on city big data, it is characterised in that including:
(1) sample data is extracted from the big data of city, the sample data includes:Bus IC-card data, bus GPS number
According to, GPS data from taxi, city POI static datas and urban meteorological data;
(2) according to the bus IC-card data on every public bus network, the bus travel amount on every public bus network is calculated,
And analysis is associated to sample data, feature set related to bus travel amount on every public bus network is extracted, by public affairs
Automobile travel amount and feature set are used as input vector altogether;
(3) unsupervised pre-training is carried out to forecast model using input vector and there is supervision to finely tune, the prediction mould trained
Type;
(4) bus IC-card data, bus GPS data, GPS data from taxi, the city POI static state of circuit to be predicted are gathered
Data and urban meteorological data, extract the feature set of circuit to be predicted, by the feature set input of circuit to be predicted train it is pre-
Model is surveyed, the bus travel amount of circuit to be predicted is obtained.
2. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 1, it is characterised in that
The step (1) also includes cleaning redundancy in sample data and incomplete data.
3. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 1, it is characterised in that
The specific implementation for being associated analysis in the step (2) to sample data is:With each bus station on public bus network
Centered on take an influence area, utilize sample data to analyze moving characteristic, geographical feature and meteorology in each influence area special
The incidence relation between resident trip is levied, feature set is obtained.
4. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 1, it is characterised in that
The step (3) includes:
(3-1) forecast model includes coding layer and logistic regression layer, and feature set and bus travel amount are the defeated of forecast model
Incoming vector, bus trip premeasuring is the output vector of forecast model;In coding layer, carried out using input vector unsupervised
Pre-training, obtains the high-rise expression of feature set;
(3-2) represents to obtain bus trip premeasuring in logistic regression layer, high-rise using feature set, using input to
The parameter of amount, bus trip premeasuring and forecast model sets up loss function, and coding layer is used for adjusting prediction model parameterses
Loss function is minimized, progress has supervision to finely tune, the forecast model trained.
5. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 4, it is characterised in that
The specific implementation of the step (3-1) is:Forecast model includes coding layer and logistic regression layer, feature set and bus
Travel amount is the input vector of forecast model, and bus trip premeasuring is the output vector of forecast model;Coding layer is by N layers
Denoising autocoder is constituted, and the reconstructed error of every layer of input vector is minimized by reconstructing denoising autocoder, and then right
Every layer of denoising autocoder carries out unsupervised training, obtains the high-rise expression of feature set.
6. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 5, it is characterised in that
The coding layer is made up of 4 layers of denoising autocoder.
7. a kind of method that bus travel amount is predicted based on city big data as claimed in claim 4, it is characterised in that
The specific implementation of the step (3-2) is:Logical layer is mainly made up of a logistic regression supervised learning algorithm, by feature
The high-rise of collection represents input logic layer, obtains the bus trip premeasuring y on i-th line road(i), gone out using bus
Input vector x in row premeasuring and i-th public bus network(i), set up loss function:
<mrow>
<mi>N</mi>
<mi>L</mi>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>&beta;</mi>
<mo>,</mo>
<msup>
<mi>D</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mo>|</mo>
<msup>
<mi>D</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>|</mo>
</mrow>
</munderover>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mi>P</mi>
<mo>(</mo>
<mrow>
<mi>Y</mi>
<mo>=</mo>
<msup>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>|</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>,</mo>
<mi>&beta;</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
1
Wherein, β is prediction model parameterses, D(s)The set of input vector is represented, P represents that is used for a function for predictablity rate,
Y represents the set of bus trip premeasuring;
Coding layer minimizes loss function for adjusting prediction model parameterses, and progress has supervision to finely tune, and what is trained is pre-
Survey model.
8. a kind of system that bus travel amount is predicted based on city big data, it is characterised in that including:
Sample data module is extracted, for extracting sample data from the big data of city, the sample data includes:Bus IC
Card data, bus GPS data, GPS data from taxi, city POI static datas and urban meteorological data;
Input vector module is extracted, for according to the bus IC-card data on i-th public bus network, calculating i-th public transport line
Bus travel amount on road, and analysis is associated to sample data, extract and go out on i-th public bus network with bus
The related feature set of row amount, regard bus travel amount and feature set as input vector;
Forecast model module is trained, for carrying out unsupervised pre-training to forecast model using input vector and thering is supervision to finely tune,
The forecast model trained;
Module is predicted, for gathering the bus IC-card data of circuit to be predicted, bus GPS data, taxi GPS numbers
According to, city POI static datas and urban meteorological data, the feature set of circuit to be predicted is extracted, by the feature set of circuit to be predicted
The forecast model trained is inputted, the bus travel amount of circuit to be predicted is obtained.
9. a kind of system that bus travel amount is predicted based on city big data as claimed in claim 8, it is characterised in that
The training forecast model module includes:
Unsupervised pre-training submodule, for carrying out unsupervised pre-training, forecast model includes coding layer and logistic regression layer, special
Collection and bus travel amount be forecast model input vector, bus trip premeasuring for forecast model output to
Amount;In coding layer, unsupervised pre-training is carried out using input vector, the high-rise expression of feature set is obtained;
There is supervision fine setting submodule, in logistic regression layer, obtaining bus trip using the high-rise expression of feature set pre-
Measurement, loss function is set up using the parameter of input vector, bus trip premeasuring and forecast model, and coding layer is used for adjusting
Section prediction model parameterses minimize loss function, and progress has supervision to finely tune, the forecast model trained.
10. a kind of system that bus travel amount is predicted based on city big data as claimed in claim 8, its feature is existed
In the specific executive mode for having supervision to finely tune submodule is:Logical layer is main by a logistic regression supervised learning algorithm
Constitute, high-rise by feature set represents input logic layer, obtain the bus trip premeasuring y on i-th line road(i), utilize
The input vector x that bus is gone on a journey on premeasuring and i-th public bus network(i), set up loss function:
<mrow>
<mi>N</mi>
<mi>L</mi>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>&beta;</mi>
<mo>,</mo>
<msup>
<mi>D</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mo>|</mo>
<msup>
<mi>D</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>|</mo>
</mrow>
</munderover>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mi>P</mi>
<mo>(</mo>
<mrow>
<mi>Y</mi>
<mo>=</mo>
<msup>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>|</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>,</mo>
<mi>&beta;</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
Wherein, β is prediction model parameterses, D(s)The set of input vector is represented, P represents that is used for a function for predictablity rate,
Y represents the set of bus trip premeasuring;
Coding layer minimizes loss function for adjusting prediction model parameterses, and progress has supervision to finely tune, and what is trained is pre-
Survey model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710324534.3A CN107194505B (en) | 2017-05-10 | 2017-05-10 | Method and system for predicting bus traffic based on urban big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710324534.3A CN107194505B (en) | 2017-05-10 | 2017-05-10 | Method and system for predicting bus traffic based on urban big data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107194505A true CN107194505A (en) | 2017-09-22 |
CN107194505B CN107194505B (en) | 2020-05-19 |
Family
ID=59872869
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710324534.3A Active CN107194505B (en) | 2017-05-10 | 2017-05-10 | Method and system for predicting bus traffic based on urban big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107194505B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107944674A (en) * | 2017-11-10 | 2018-04-20 | 华中科技大学 | A kind of method using on-line off-line data assessment commercial planning |
CN108182196A (en) * | 2017-11-27 | 2018-06-19 | 东南大学 | A kind of Urban traffic demand Forecasting Methodology based on POI |
CN109191846A (en) * | 2018-10-12 | 2019-01-11 | 国网浙江省电力有限公司温州供电公司 | A kind of traffic trip method for predicting |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1407513A (en) * | 2001-09-13 | 2003-04-02 | 孔勇 | Monitoring bus punctual rate system |
CN105243444A (en) * | 2015-10-09 | 2016-01-13 | 杭州尚青科技有限公司 | City monitoring station air quality prediction method based on online multi-core regression |
CN106295906A (en) * | 2016-08-22 | 2017-01-04 | 万马联合新能源投资有限公司 | A kind of urban public bus lines querying method based on degree of depth study |
-
2017
- 2017-05-10 CN CN201710324534.3A patent/CN107194505B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1407513A (en) * | 2001-09-13 | 2003-04-02 | 孔勇 | Monitoring bus punctual rate system |
CN105243444A (en) * | 2015-10-09 | 2016-01-13 | 杭州尚青科技有限公司 | City monitoring station air quality prediction method based on online multi-core regression |
CN106295906A (en) * | 2016-08-22 | 2017-01-04 | 万马联合新能源投资有限公司 | A kind of urban public bus lines querying method based on degree of depth study |
Non-Patent Citations (2)
Title |
---|
任其亮: "时空路网交通拥堵预测与疏导决策方法研究", 《中国博士学位论文全文数据库 工程科技Ⅱ辑》 * |
潘若愚: "物联网环境下基于改进蚁群的大城市公交快速响应、运能优化与评价研究", 《中国博士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107944674A (en) * | 2017-11-10 | 2018-04-20 | 华中科技大学 | A kind of method using on-line off-line data assessment commercial planning |
CN108182196A (en) * | 2017-11-27 | 2018-06-19 | 东南大学 | A kind of Urban traffic demand Forecasting Methodology based on POI |
CN108182196B (en) * | 2017-11-27 | 2021-09-07 | 东南大学 | Urban traffic demand prediction method based on POI |
CN109191846A (en) * | 2018-10-12 | 2019-01-11 | 国网浙江省电力有限公司温州供电公司 | A kind of traffic trip method for predicting |
Also Published As
Publication number | Publication date |
---|---|
CN107194505B (en) | 2020-05-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105206048B (en) | A kind of city dweller colony transfer mode based on traffic OD data finds system and method | |
CN103984994B (en) | Method for predicting urban rail transit passenger flow peak duration | |
CN110836675B (en) | Decision tree-based automatic driving search decision method | |
Zhao et al. | Identification of land-use characteristics using bicycle sharing data: A deep learning approach | |
Yu et al. | Extracting and predicting taxi hotspots in spatiotemporal dimensions using conditional generative adversarial neural networks | |
WO2023109720A1 (en) | Urban public transportation planning method | |
CN105868861A (en) | Bus passenger flow evolution analysis method based on time-space data fusion | |
CN107392389A (en) | Taxi dispatching processing method based on ARIMA models | |
CN103116702A (en) | Bicycle-mode traveling selection forecasting method based on activity chain mode | |
CN107194505A (en) | A kind of method and system that bus travel amount is predicted based on city big data | |
CN111931998B (en) | Individual travel mode prediction method and system based on mobile positioning data | |
CN110070255A (en) | The method for introducing commuter's Passenger Traveling Choice modeling and analysis after sharing bicycle | |
CN106021902A (en) | Grading method used for urban rail transit passenger flow congestion index | |
CN113344240A (en) | Shared bicycle flow prediction method and system | |
CN115204477A (en) | Bicycle flow prediction method of context awareness graph recursive network | |
Rasaizadi et al. | Short-term prediction of traffic state for a rural road applying ensemble learning process | |
CN111008730B (en) | Crowd concentration prediction model construction method and device based on urban space structure | |
CN113159371A (en) | Unknown target feature modeling and demand prediction method based on cross-modal data fusion | |
Shao et al. | Estimation of urban travel time with sparse traffic surveillance data | |
CN110610446A (en) | County town classification method based on two-step clustering thought | |
Manjushree et al. | Household Vehicle Ownership Prediction Using Machine Learning Approach | |
Lei et al. | Research on multi-objective bus route planning model based on taxi GPS data | |
CN109523146B (en) | Scheduling region division method based on complex network | |
CN115600710A (en) | Method for predicting regional demand of manned vehicle, electronic device and storage medium | |
Cheng et al. | Method for predicting travel time of motor vehicle based on stack noise reduction self-encoder |
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 |