In order to place the best location for the new proposed Rubsains supermarket, I will necessitate to analyze current methods used by bing supermarkets ( and retail merchants in general ) . From at that place, I will be able to use some of these techniques to bring forth a list of standards, which will assist me contract down the possible locations.
Clarkson, Clarke-Hill and Robinson, 1996, sought to look into the bing location choice processs of 16 different food market retail merchants, by supplying them with a questionnaire. They found five different techniques used by the retail merchants. I will really briefly explicate these below:
Checklist analysis – list of factors on population, handiness, competition and cost.
Analogue method – Birkin, G. Clarke, M. Clarke, 2002, explain this as
“ Attempts to calculate the possible gross revenues of a new ( or bing ) shop by pulling comparings ( or analogies ) with other shops in the corporate concatenation that are likewise in physical, locational and trade country fortunes. ”
Obviously, the success relies on being able to happen these “ similar ” sites.
Clarkson, Clarke-Hill and Robinson, 1996, suggests the chief countries of survey include ; “ client studies, competitory analysis, and the aggregation of demographic and socio-economic information. ”
Fiscal analysis – concerned with site acquisition costs, and operational costs vs. likely turnover. It is referred to as a “ complementary process ” .
Arrested development patterning – Birkin, G. Clarke, M. Clarke, 2002, explains arrested development as:
“ Specifying a dependant variable, such as shop turnover and trying to correlate this with a set of independent or explanatory variables. Coefficients are calculated to burden the importance of each independent variable in explicating the fluctuation in the set of dependent variables. ”
However, arrested development patterning “ evaluates sites in isolation, without sing the full impacts of the competition ” . There is besides an premise that the variables are independent, when this is non ever the instance. However, the most of import factor they identify is that “ arrested development theoretical accounts fail to manage adequately spacial interactions or client flows ” .
Gravity patterning – based on spacial interaction theory. Rogers, 1992, states that
“ Gravity theoretical accounts forecast shop gross revenues through the coincident consideration of retail attraction, i.e. sizes and gross revenues, and the distance relationships between viing installations and population distribution and denseness. ”
Simply speech production, if a shopping Centre is every bit accessible, the more attractive one is chosen. When the Centres are every bit attractive, the nearest Centre is preferred. However, as Birkin, G. Clarke, M. Clarke, 2002 point out, “ these penchants are non deterministic ” .
Overall, it was found that the retail merchants questioned in the probe used a mixture of these theoretical accounts, instead than a choice of one specific attack.
For illustration, checklist analysis allows the retail merchant to analyze many sites. This means that merely the most suited locations are concentrated on. From at that place, the linear attack can be used. Customer interviews provide the retail merchant with information about the “ beginnings of the clients, passing features, demographic composing, shopping behavior and media usage wonts ” . You can besides cipher how long it takes to go ( via auto or public conveyance ) to rivals from the proposed site location. The rival shops are assessed in footings of “ visual aspect, ware presentation, quality, location and service degrees ” . Information can be retrieved on client demographics in the country where the site is expected to be. Tendencies can be identified by comparing this information with in-store informations, for illustration, general purchase wonts. As a consequence of this, it will “ enable the retail merchant to gauge the possible market portion in the shop ‘s catchment country and possible returns ” . ( Clarkson, Clarke-Hill and Robinson, 1996 ) .
Rogers, 1992, suggests that the cardinal elements to see ( regardless of the theoretical account of analysis used ) can be highlighted in the undermentioned tabular array:
Table from: “ Retail Site Analysis is Much More Than “ Location, Location, Location ”
The first component, “ Site Characteristics ” , was broken down into several factors. These are outlined below:
Alone Selling Point ( USP )
The USP for Rubsains is the “ organic merchandise scope sourced from local farms ” . There are a twosome of factors that need to be considered in order to carry through this USP. First, there is the fiscal issue of transporting the green goods from the farms to the supermarket. In peculiar, the cost of gasoline will increase the farther off the supermarket is from the farms. It is besides of import to see the spoilage of merchandise when transported over larger distances. By and large in the UK, distances are non so huge so it is non a immense job, but evidently the closer the supermarket is to the farm, the freshman the green goods will be.
Rubsains has a loyal client base, as a consequence of the determination to accept a lower net income border than other supermarkets. This trueness could be backed up by the gravitation theoretical account discussed in the old subdivision. The clients are more “ attracted ” to this supermarket, so may be more willing to go greater distances than to more local supermarkets, which do non hold this trueness. This would propose that puting the supermarket a small farther off from the mark demographic is non so much of an issue. It will, nevertheless, need to be weighted among other factors. It is besides of import to see the general demographics. Rubsains higher monetary value scopes will pull a more affluent demographic, so will non profit from poorer countries.
Conveyance and Environment
Like many supermarkets, Rubsains have implemented policies to assist the environment. In peculiar, they encourage the usage of public conveyance. Therefore, the supermarket should be near to develop and bus Stationss to endorse up their committedness. However, they do supply a free coach service to cardinal locations. This suggests that puting the supermarket near public conveyance should hold a high weight, but is non indispensable. A auto park is besides provided for people who want to drive in, though this is non encouraged. Therefore, it may be worth merely sing a little size auto park. This will besides assist in footings of fiscal costs.
Rogers, 1992, mentioned the thought of go throughing traffic. For those who are driving to the supermarket, the sum of traffic on the route needs to be considered. Too small traffic and the supermarket does non profit from excess clients. Excessively much traffic and the clients will be put off. Customers may besides be put off from utilizing paths with big new waves, such as on a expressway.
Visibility is another strong factor to see, thanks to the information provided by Rogers, 1992. How far off should the supermarket be seen? What way can the supermarket be seen at? What are the impacts of other objects in position? For illustration, are at that place other edifices or trees in the manner?
As a consequence of this analysis, I have produced a list of standards to see:
Supermarket must be as close to farm land as possible, to keep fresh green goods
Supermarket must be placed in a comparatively affluent country, above mean income.
Supermarket must be as close to bus/train Stationss as possible for easiness of entree
Supermarket must be near to chief roads with moderate traffic, but non expresswaies
Supermarket should be easy seeable, by understating objects in the manner
Supermarket should non be built on steep inclines, no more than 2 grades.
Supermarket does non necessitate huge infinite ( e.g. for auto park ) , in order to cut down fiscal costs
I imported the net income study informations from the Wiki page and converted the vector points into a raster map. In order to make full all raster cells with income informations, I used spline based insertion. Due to the limited figure of informations points, I smoothed out the algorithm so that I was non excessively limited on the figure of “ affluent ” musca volitanss.
& gt ; v.to.rast input=income output=income use=z
& gt ; v.surf.rst input=income elev=incomegrid smooth=30 tension=70
I used r.reclass to sort all cells with values greater than 29999 to 1, otherwise 0.
Figure: Interpolation Figure: Reclassification
Bus Stop/Train Station Data
In order to acquire local conveyance informations ( i.e. for coach Michigans and train Stationss ) , I used the Transport bed of OpenStreetMap. By utilizing the Overpass API, I was able to pull out the coach stop/train station vector informations.
The question involved stipulating some properties to filtrate the information on, including the type of informations ( e.g. “ bus_stop ” ) and a boundary box for consequences. To acquire the box coordinates, I took the top-left and bottom-right co-ordinates of the Leicestershire country and converted them from Ordnance Survey ‘s National Grid coordinates to WGS84 lat/lon co-ordinates. Below is the full question for the coach stops informations:
& lt ; osm-script & gt ;
& lt ; query type= ” node ” & gt ;
& lt ; has-kv k= ” main road ” v= ” bus stop ” / & gt ;
& lt ; bbox-query e= ” -1.1730235 ” n= ” 52.793566 ” s= ” 52.685014 ”
w= ” -1.3492839 ” / & gt ;
& lt ; /query & gt ;
& lt ; print/ & gt ;
& lt ; /osm-script & gt ;
The end product of the question provided me with an XML file with all the points, among other properties for nodes. In order to acquire this into GRASS GIS, I imported the informations utilizing v.in.ogr. Since the co-ordinates were still in the WGS84 lat/lon system, I had to click “ Override dataset projection ( use location ‘s projection ) ” .
I converted the co-ordinates into the current projection utilizing the undermentioned bid in the shell terminus.
& gt ; v.out.ascii waypoints | m.proj -i
Figure 3: Converting co-ordinates to current projection
Finally, all I needed to make is re-import the converted information, this clip utilizing v.in.ascii.
The route informations from the dataset provided in labs did non line up with the coach informations, so I decided to import new route informations from OpenStreetMap. I used a similar question to the coach halt one from earlier. The chief difference was the information end product. The XML information is no longer a series independent node co-ordinates. Alternatively, it contains a muddled up list of point nodes from all the roads. Below that are the existent route nodes, which reference the relevant point nodes from above ( many are re-used ) . In order to acquire this information into GRASS GIS utilizing v.in.lines, it needed to be in a specific format where, for each route, you list all the co-ordinates that make up the route, followed by “ NaN NaN NaN ” ( terminal of route ) . Using jQuery, I was able to re-format this. The following job was the fact that m.proj -i can non cover with the “ NaN NaN NaN ” twine, and since we need to maintain path of this to separate between each route, I needed to utilize some shell bids to acquire around this.
& gt ; cat roads.txt | sed s/NaN/0/g | m.proj -i | sed “ s/62
2695.72.+/NaN NaN NaN/g ” & gt ; out.txt
This codification replaces the “ NaN NaN NaN ” with 0 before the transition. At that phase, it will try to change over 0. Since this value will be the same each clip, it is easy to descry. All that is left to make is replace those converted values back. Finally, it was the simple undertaking of importing the line informations with v.in.lines.
Once the route information was in GRASS GIS, I added a 300m buffer around the roads utilizing r.buffer
The new route informations showed the orbiter image to be even more inaccurate than it was with merely the standard route bed from the lab dataset. To acquire around this, I decided to draw a new satellite image from Google Maps and execute geo-referencing to place the co-ordinates right. This was done utilizing ArcMap, as it provides a somewhat more synergistic method. This was clip devouring but worth it. The new imported map has a better alliance.
In order to avoid trees that could “ conceal ” the supermarket, I decided to make a negative buffer around forest countries. I besides needed to avoid preies and H2O. I used r.reclass to bring forth a new bed, with merely the forest, prey and H2O information as 1, and everything else as NULL.
& gt ; r.reclass input=landcov output=bad_land
& gt ; 3 = 1
& gt ; 4 = 1
& gt ; 8 = 1
& gt ; * = NULL
In order to acquire the negative buffer, I foremost had to make a 300m buffer around the country.
& gt ; r.buffer bad_land distance=300 output=bad_land_buf
Following, I used r.mapcalc to trade the 1 and 0 values around.
& gt ; r.mapcalc
& gt ; bad_land_negbuf = if ( isnull ( bad_land_buf ) ,1,0 )
Figure 4: 300m buffer around “ bad ” musca volitanss Figure 5: Negative buffer
My research did non cover the job of incline surfaces in the arrangement of supermarkets, but it is something of import to see. Using the leisure centre illustration from the lab sheet, I was able to utilize the correspondent theory ( as described earlier ) with the fact that a leisure Centre is likely traveling to hold similar standards to a supermarket in footings of incline surfaces. It is a strong premise, but it is the best I can come up with given the sum of research done.
Finally, I combined my standards beds to happen locations which match. I have made the premise that you can construct on arable/pasture land since my research did non cover the types of land you can construct on.
& gt ; r.mapcalc
& gt ; possible_sites = if ( isnull ( bad_land_negbuf ) ,0,1 ) *
if ( isnull ( ab_roads_buf ) ,0,1 ) *
if ( isnull ( bus_stops_buf ) ,0,1 ) * level * wealthy
Figure 6: Concluding solution ( two possible locations in circle )
You can see from the top left corner that there are two locations where the supermarket can be built. Puting the co-ordinates of the possible locations into a map service, it came up with a moderate sized small town named Belton.
Note: there are three bus Michigans hidden underneath the two proposed supermarket locations.
There have been a figure of premises and estimates made in this undertaking. The first thing to observe was the inaccuracy of the informations provided from the labs. This became evident when I plotted the coach halt and route informations from OpenStreetMap. This unfastened beginning informations chiefly comes from authorities and commercial beginnings, so one could presume that this information is likely to be reasonably accurate. However, harmonizing to the OpenStreetMap wiki page ( see mention ) , they rely on many voluntaries to roll up the informations with portable GPS devices. We do non cognize how accurate these measurings are, and how dependable the voluntaries are.
Harmonizing to Lucy Bastin, 2013, the satellite image from the labs came from a 1984 dataset. GIS was non peculiarly advanced back so, with old engineering and techniques used to roll up informations. The Bartholomew dataset is besides dated ( from 1995 ) , and went through a batch of transitions and processing. As is frequently the instance, the more you convert informations, the less accurate and precise it becomes. There is no surprise here that both the orbiter image and the dataset are out of sync. It is interesting to observe the difference in the truth from the 1984 orbiter image, to the 1995 dataset, to the reasonably recent OpenStreetMap information. This clearly shows how far GIS has come in the last decennary.
Since the new route and coach informations was in a really different place from the orbiter image, I had to draw a new image from Google Maps and execute geo-referencing. This procedure was a small arduous, but I was able to obtain a reasonably accurate map. Not every portion lined up absolutely, but given the clip restraints, and the fact it was merely traveling to be used for a background, it was accurate plenty for the undertaking.
Further premises were made in footings of the land screen and wealth informations. I decided that the procedure of acquiring new land screen informations would be excessively clip consuming, so I decided to utilize the Bartholomew dataset. It is rather possible that land screen I have assumed for my concluding consequence is incorrect. My supermarket may be submerging in the center of a reservoir.
The wealth informations was provided on this faculty ‘s wiki page. However, there is no meta informations to state me where, when and how it was collected. Equally good as this, the figure of informations points is rather minimum. This became a job with the insertion of the information. The less data points you have, the more premises are made during the procedure. For illustration, one individual with high income converts to a big affluent country. It is possible that a individual affluent individual is populating in a lower category country. The smooth/tension factor will evidently modulate this to an extent.
I placed a 200m buffer around all the coach halt points, presuming that people can walk that distance from the coach halt to the supermarket. This is an about 2-3 min walk, which is sensible for most people. However, for more aged people, this may be a job. The buffer besides does non take into history the figure of bus Michigans. There is no point puting a supermarket on a street with one coach halt. Some kind of constellating algorithm would assist to better this. In my solution there are three bus Michigans ( near to each other ) by the proposed locations. In this instance, it is non excessively bad, since these coach Michigans are merely two stat mis off from a big bunch in Shepshed, so there are plentifulness of conveyance links.
Overall, I think GIS in recent old ages is going really utile and efficient at gauging where something should be located. However, it does to a great extent trust on the informations being accurate in the first topographic point. This can be improved by guaranting that those who create the informations include meta informations, so that users can construe and make up one’s mind whether to swear and utilize it. The sum of information is besides an of import factor. It does non count how accurate the information is, if there is non plenty of it, the GIS techniques will non be good plenty to generalize the information. This is surely the instance with insertion, as pointed out earlier with the income informations. The tools in GRASS GIS are reasonably basic, and I believe there are better tools and algorithms that aid to better the efficiency of GIS undertakings, and possibly better trade with inaccuracies in the information. Some of the research documents I looked at did propose that some supermarkets in the yesteryear were loath to trust on GIS, and frequently used more traditional techniques. However, it looks like there is turning demand for it these yearss.