AU2018101529A4 - A device of evaluating the dewing conditions of a house - Google Patents

Abstract

Abstract With the development of the city and the improvement of the living standard, people pay more and more attention to higher-level dwelling and traffic conditions around their houses. However, the existing housing-evaluation products cover limited information, by which people cannot evaluate certain living locations in an immediate, comprehensive and objective way. It may cause confusion or waste massive time when they purchase or rent a house. The purpose of this patent is to provide tenants a dwelling-evaluation service, by which they can choose their favorite houses efficiently. The service covers four indicators: the residential-environment indicator, the travelling-convenience indicator, the living-convenience indicator and the living-safety indicator that are calculated based on nine types of Big Data or remote sensing images. They can evaluate the living conditions in a micro-level and also improve the accuracy and efficiency of the current housing-evaluation products. Data types Distribution of vegetation, water and garbage Remote residentnal sensing environment image data , indicator Fundamental Service facility pattern information geographic living Tenement and service facility convenience spatial distribution indicator Social housing data travelling Public safety incidents spatial distribution convenience indicator Taxi data Space-time distribution of subway congestion living-safety Metro card indicator data Distribution of designated taxi services Fig. 1 ndicatoi e culai Data preprocessing Fes n-]yn indicator p Iuri Text Housing troen s- fndioatsc livabilitymdnaorincato processing assessment / ndicator Pines residential i cc evironmnent indicator i,, to h,,-I Fig.2

Description

TITLE A device of evaluating the dewing conditions of a house

FIELD OF THE INVENTION

The patent is a device used to evaluate the livability of a house scientifically and objectively. The patent collects data from nine kinds of sources and figures out four indicators to evaluate the house conditions. It provides accurate, immediate and comprehensive information of the houses to help tenants to better choose their satisfactory houses.

BACKGROUND OF THE INVENTION

With the rapid development of the urban infrastructure and the economy industry, more and more transient population flocks to the city. The increasing floating population promote the development of local economy and commerce, which makes the quality of life soaring.

However, this kind of phenomenon also brings a series of problems such as the rising of housing costs, the deterioration of the environment, and the travel costs rising caused by the increasingly traffic congestion. More importantly, security problems happened frequently due to the excessive pedestrian volume. Public resources, especially the resource of medical and education was extremely tense.

On the premise of sustained economic development, and people ’ s living standards increasing, the requirement of their life quality and living conditions has become higher and higher. They began to pay attention to their health and environment, and they realized community environment has a closely relationship between the citizen’ s health and life comfort.

In fact, livability which as the evaluation standard of “Human Settlement” has widely recognized in these days. Moreover, livability as the essential aspect of “Human Settlement” scientific research has become a research focus. However, researches of the city’ s livability questions are not a new question. Actually, United Nations started “UN Habitat Award” from 1989. At the same time, E. Howard, C. A. Doxiadis, P. Geddes, L. Mumford as a representative of urban planning pioneers created the study of "human settlements".

Nowadays, with developing of cities, the rental market increase rapidly because of more and more migrants. However, the immigrant population lacks the understanding of the community’s environment, the distribution of public resources and the current situation of transportation.

Additionally, it’ s hard to find a helpful and authoritative information source, and so, a leasing APP that can provide objective and comprehensive evaluation is extremely important.

Although there are varieties of current housing-evaluation products or agencies, the main information source of these products are basically from homeowners, in addition, those information platform can only cover the most basic information, such as location, rent, structure, but the other important information are not involved.

Firstly, lack of multi-modal travel information, which means most renters do not know the exact time they need and whether it is comfortable in different travelling methods. Moreover, lack of travel information in complex situations make tenants cannot obtain information about travel conditions at different periods. More importantly lack of residential-environment understanding means it is difficult for tenants to have a comprehensive understanding of the security, convenience and surrounding ecological environment of the house.

The result is that the consumer bought or rent a house but not satisfy with the housing conditions, eventually, this phenomenon might lead to the unbalanced usage of a large number of public facilities. Which means, some areas are under excessive pressure, while, in other areas, there may be a phenomenon that the waste of resources and the public facilities are not fully utilized.

Combined with these current problems, we use trace matching, spatial co-location mining, deep learning and other technologies along with the geographical information and remote sensing data. Actually, it is the first time that remote sensing technology has been introduced into the field of housing leasing and saling. and this kind of method can improve the personalization and usability of service.

SUMMARY OF THE INVENTION

As the lack of effective assessment tools to evaluate the comprehensive indicator of a house, the patent aims to combine different kinds of data from nine sources and figures out four indicators (the residential-environment indicator, living-convenience indicator, travelling-convenience indicator, living-safety indicator; show in Fig.l ) based on big data or remote sensing images. The evaluation method is shown in Fig.2. Through the rich data source and the scientific algorithm, the assessment is much more accurate and reliable than most of the products on the market. The invention aims to evaluate the housing resources by four indicators: 1. Travelling-convenience indicator ( shows in Fig.3 ):

For most office workers, the convenience of travel is always the first factor they will consider. Thus, it’ s quite necessary to evaluate users ’ travelling-convenience indicator. The invention extracts user’ s work place and living place through analyzing taxi data, subway data and public roads data. The travelling-convenience indicator is evaluated by the average time and standard deviation of transit time of travel consuming. 2. Living-convenience indicator ( shows in Fig.4 ):

Shopping, catering and medical care are also important factors influencing the choice of residence. Therefore, it is necessary for us to evaluate the convenience of residential areas which is based on the relevant POI data. We use weighted analysis and co-location pattern mining to comprehensively evaluate the convenience of housing life. 3. Residential-environment indicator ( shows in Fig.5 ):

The Residential-environment is an indispensable factor in people's choice of residence. We obtain China's Gaofen-1 satellite data and

Google Earth HD remote sensing image data to extract information about surrounding environment of the house ( for example, vegetation area, water area, garbage area, etc.). Thus, we can carry out a reliable assessment about the housing residential-environment. 4.Living-safety indicator ( shows in Fig.6 ):

Apart from the three indicators above, living-safety indicator is another important factor to evaluate the dwelling conditions of a house. Machine learning is used to analyze the evaluations of house from websites. Through feature selection and weighting method, an emotional features dictionary for weighting different features is built. Based on machine learning, we score the different safety misadventures within a 3km buffer area around the house. The safety misadventure will be scored from 1 to 10. 10 is very dangerous while 1 is minor security risks. At last the living-security indicator is figured out by adding different scores.

Description of drawings

Fig.l is the sources of data and the four indicators to evaluate the evaluate the dewing conditions of a house

Fig.2 is the data processing of four indicators

Fig.3 is the calculating process of travelling-convenience indicator. It explains how to deal with the text information and how the indicator is figured out.

Fig.4 is the calculating process of living-convenience indicator. It shows the process of data processing and indicator calculation.

Fig.5 is the calculating process of green coverage rate, water coverage rate and regional waste index. The residential-environment indicator is calculate with these three data.

Fig.6 is the calculating process of living-safety indicator. It shows how text information is dealt with and how the indicator is calculated.

Fig.7 is the design mentality of “FE ZU” application

Fig.8 is a GUI of “FE ZU” application

Fig.9 is a GUI of “FE ZU” application

Fig. 10 is a flowchart of machine learning

DESCRIPTION OF PREFERRED EMBODIMENT 1. Travelling-convenience indicator:

For the largest amount of office workers, the convenience of travel is always the first factor they will consider. Thus, it’s quite necessary to evaluate users’ travelling-convenience and comfort indicator. The invention extracts user’s work place and living place through analyzing taxi data combined with public roads and subway data.

Data processing: 1) When dealing with the data, we eliminate the invalid taxi data at first. Then, we match user’s getting on and off position to the map. 2) Determining the type of travel, calculate the travel time 3) Calculating the travel time and travel stability of the house according to the commute time set by the user, filter out the record pairs to meet the conditions. 4) Ascending housing sources by average travel time consuming and standard deviation consuming. 5) The shorter the average time, the easier it is to travel; The less time-consuming standard deviation, the better the travel time stability. 2. Living-convenience indicator:

Shopping, catering and medical care are also important factors influencing the choice of residence. Therefore, it is necessary for us to evaluate the convenience of residential areas which is based on the relevant POI data. We use weighted analysis and co-location pattern mining to comprehensively evaluate the convenience of housing life.

Data processing: 1) Building a standard tree of service circles, which is based on POI data types. 2) Extracting the importance weight of the typical feature types around the house based on the POI data. 3) On the basis of weighted analysis, the weights obtained from the living-convenience indicator are calculated by the quarter-clock service range of a specific house, and the presence or absence of relevant types of feature points within the service range is counted. 3. Residential-environment indicator:

The Residential-environment is an indispensable factor in people's choice of residence. We obtain China's Gaofen-1 satellite data and Google Earth HD remote sensing image data to extract information about surrounding environment of the house (for example, vegetation area, water area, garbage area, etc.). Thus, we can carry out a reliable assessment about the housing residential-environment.

Data processing: 1) Use the NDVI algorithm to figure out the vegetation coverage rate.

The calculation formula is as follows:

NIR + R

In the formula, NIR and R represent the reflection values of the near-infrared band of the pixel and the reflection values of the red band, respectively.

Secondly, the classification threshold is set by expert scoring, and the image is classified into a vegetation area and a non-vegetation area. This product uses 0.7 as the threshold. When NDVI is greater than 0.7, the pixel represents green space. The formula is as follows:

2) Use the band reflectance to calculate the water area, and calculate the formula as follows:

In the formula, Green is the reflection value of the green band, and SWIR is the reflection value of the short-wave infrared band. 3) Calculate the garbage indicator. After constructing the garbage sample data set, the garbage dump target is detected through machine learning, and then the regional garbage indicator is calculated.

The specific learning process is as Fig. 10.

We use this well-trained deep learning CNN model to detect garbage

dump targets in cities. According to the test results of the model, the area of the garbage dump is counted. Focusing on the 3km buffer analysis of each housing interest point, the number and area of garbage points around different houses are counted, and then the environmental indicators based on garbage dump are obtained.

Calculation of residential-environment indicators:

On the basis of the above three aspects, the expert scoring method is adopted to set the corresponding weight for each sub-indicator as the residential-environment indicator:

4. Safety indicator 1) Text pre-processing

Encode the file specifications, the format of the corpus stored download from the internet can be very different, which is very disturbing to the experiment. So the first step is normalization processing to the corpus data format. 2) Feature Selection and weighting

After pre-processing and formalizing the text, we get a feature space, the number of features can reach tens of thousands of dimensions or even

hundreds of thousands of dimensions, which not only makes the computation time longer, but also reduces the accuracy of classification to a large extent. By using the unique classification method of feature selection, we can avoid the problem, and improve the classification accuracy at the same time compared with the same type product promotion efficiency. In addition, we have built an emotional features dictionary in the security field and base on the dictionary to feature weight assets, feature with strong representation ability is given greater weighted, and the feature with weaker classification ability is given a smaller weight, which can effectively restrain the noise. 3) Security classification

Classifier is the core of text classification problem,and this product mainly uses to support vector machine to classify and complete the operation by existing Lib SVM. Than,the obtained security-related will be graded from 1 to 10. 10 is very dangerous while 1 is slight security risks. 4) Geographic information extraction and geo-coding

Extract the locations where security-related events occur in public information and use Baidu's geocoding API to geocode. We can get the latitude and longitude of the incident, establish the Security event library. 5) Calculation of the security indicator housing resource

Set a 3km buffer zone for each housing resource and extracts the security events that occur within the buffer zone, sum up all security score and get total safety hazard indicator.

Example 1

The living ability evaluation method is used in a housing-evaluation product called “LE ZU”. The app is made up with 6 different components. Specifically, four of the components are based on the different evaluation indicators from our method. 1) Housing resource information Statistics Inquiry module:

In order to meet the needs of users to choose their own living conditions, we through a large data management system to provide users the housing resouce information with conveniently and efficiently. After users enter or select their own personalized requirements (for example, room type, price and room, etc.), the system through the search database, to provide users with the availability of information to meet their needs (including, location, price, landlord and house contact details, etc.). 2) Electronic Map Call module:

Using the available bus line data, subway line data, public provincial administrative boundary map, as well as digital remote sensing image maps, in the geographical code to meet the needs of the rental system map, after geocoding to meet the needs of the rental system of electronic map,the use of GeoServer such as map release platform to publish, apps and Web pages to implement map calls. This module does not require user input. 3) travelling-convenience analysis module:

This model Calculate the travel accessibility of the user's work location to a certain residential area by analyzing Shenzhen road map, subway line data and bus line data, etc. Through this, the app can evaluate the travelling-convenience indicator for user once the user put in his or her work location. 4) Residential Residential-environment evaluation module:

This model extracts the Remote sensing images from “High Score 1 Satellite” and “High Score 2 Satellite”. From these images, it can figure out green coverage rate, water coverage rate and regional garbage indicator around the community. Based on the above three aspects, the expert scoring method is adopted to set the corresponding weight for each sub-indicator as the Residential-environment indicator. 5) Community security evaluation module:

Based on crime data, take statistical analysis of crimes in the community and surrounding areas to achieve community safety evaluation. For example, it can Crawl data from Sina Weibo and using deep learning method to filter valid text for analyzing. Thanks to the large amount of sources and scientific analyzing method, the security indicator is obviously accurate and reliable. This module does not require user input. 6) Housing convenience evaluation module:

Based on the published POI data, using the neighborhood statistical analysis method to achieve the convenience evaluation of the house. This module does not need user’s input.

Product Value

On the one hand, for social value, the public can get housing recommendations from this product, which can reduce the expenditures and traffic congestion costs. Moreover, the public transport operator can use our information to optimize the selection of existing bus routes. More importantly, governments can balance traffic supplying and demanding to ease the living pressure of large residential areas.

As for the commercial value,the demand of housing rental have a huge market, firstly, this product can make a profit from advertisements and query traffic application information, moreover, rental agent would charge small intermediation fees from this product. Due to the geografical information and Big Data, we can reprocess the data and provide decisions support for bus companies and the government.

Claims (3)

1. Claims

   1. A device of evaluating the dewing conditions of a house , in which current housing-evaluation products on the market only provide limited information such as some basic information and subjective description of the house; it is difficult for users to evaluate or select house scientifically from this information, these housing-evaluation products cannot meet users' daily needs diversely and comprehensively
2. 2. A divice as claim 1 said, in wich different groups of people demand real-time travelling information such as travel time or traffic mode, existing housing-evaluation products on the market cannot provide an immediate and comprehensive stuff of travelling from a house start to the destination in different scenarios; the patent is designed to provide different travelling information in multiple scenarios by using big data such as traffic flow data; the relevant comfortable and convenient models are designed to help people with their house selections for different travel times and traffic modes.
3. 3. A divice as claim 1 said, in wich the patent uses band reflectance of the recent remote sensing images (MODIS and Google high-resolution images) to calculate the water area, green area and dump area respectively around the house based on the NDVI algorithms and deep leaning methods; it carries out environmental assessments surround a house reliably to supersede the subjective description in the current housing-evaluation products.