Correspondence Analysis using SPSS

[Music] hello welcome to my easy statistics in this video we will be discussing about correspondence analysis correspondence analysis is one of the statistical methods in multivariate analysis correspondence analysis will allow us to examine the relationship between two nominal variables graphically in a multi-dimensional space to do this correspondence analysis I am using an example in which I have two variables nominal variables one is product the other is fragrances product variable is given in this table as a rose and fragrances are given as columns ok the researcher want to know the relationship between product and fragrance that is if a customer uses a bath so what type of fragrance he is expecting from the bathroom if a customer is using shampoo what is the preferred fragrance he likes to have in the shampoo now we are talking about the relationship between product and fragrances this table is called correspondence table where we have product and fragrances and we have some values here these values are called score or called frequency that is a customer who is have bathroom using bath so want sandal 314 members have selected that is number of people who like to have this fragrances that respondents are being entered here bath soap fragrance rose fragrance bath so product rolls wagons want 148 so in this way the frequencies are given in this table now the relationship between the product and fragrances are given in a multi-dimensional scale this is a two dimensional scale we have here we have you can see the green color is a product and the blue color is fragrances hair oil and shampoo are coming closer relationship whereas body lotion deodorant are having close relationship if we carefully observe will understand hair oil is having a better relationships with a jasmine fragrance whereas shampoo is having with the menthol if you see the body lotion body lotion has relation with the lavender and deodorant with rose and you can see yes sandal fragrance is more closer with bath so now this is what correspondence analysis will be useful for to identify the relationship between non-parametric variables are a nominal variables fragrance and product are nominal variables I'll do this example using SPSS okay this is the SPSS file where I have three variables fragrance product and score the fragrance is having values these are the six values we have and product is having six values okay score is a frequencies also the data view we can see this data you may we have product bath soap rose fragrance is 148 people want this sort of fragrance in the same way we can see shampoo with rose is 42 so the table is being entered here in three variables frequency product and core now let me start the process of analysis for correspondence to do this analysis we go for analysis analyzed in analyze we go for dimension reduction in dimension direction we have correspondence analysis okay before I do this I need to do some settings that if score is the frequency so I need to do with cases for that so before I go for correspondence analysis I want to do weight the cases because different scores are they here so let me start with data let me go to weight cases frequency and we will wait the cases by score now it is done we have waited the cases now we'll start the correspondence analysis we go for analyze dimension Direction correspondence analysis in this ok product would take as a rope Frances as column because we want to know the relationship between product and fragrances find the range product has six values one to six minimum and maximum update it continue in the same way for fragrances also we define the range 1 to 6 update it continue now this is into setting we did that we took a row as product and column as fact fragrances will go for the model in this normalization method we are using you symmetric and we have number of dimensions in solutions to we can even make it to three dimension for dimension but I will discuss in my analysis further why we are taking to as a base and the distance measurement we are taking as Chi square continue we go for plots by pole by Pole is combination of two nominal data I can have only for the row that is product and column separately say continue okay now in this window I have taken product a zero column as fragrances the model has taken dimension as true and the normalization method as to metrical plotting by plot I have taken roblox and column plot also we have taken say continue and okay okay this is the correspondence analysis will discuss one table after another let us see the first two table this table is correspondence table which are already shown in the PPT way we have product as the rows and fragrances as column so there are some various some respondents who have not answered missing values also they because all people are not uniform who have responded more people have responded for bath soap and less people have responded for hero in the same way a rose and jasmine are having high frequency of flavours whereas the lily is having very less so this is what we can basically understand by seeing the table more people have answered for bath soaps followed by shampoo very less people have 1/3 for heroine fragrance also we can see the fancy fragrances people like their mo is jasmine rose sandal very few people want lily as a fragrance okay let us go further to next table table is where we are going to see proportion of inertia this is important here we have five dimensions one two five dimensions but let us see the initial how is going to account for each dimension one after another for the first dimension if you see we have the proportion of inertia is 0.65 665 percent inertia it is showing if I go to dimension - it is increasing by 0.205 that is 25 or 20 percent increasing so the total inertia explained is 0.8 6 to go to third dimension we can see the counted for inertia is very less that is 0.09 that is 9 percent only we are getting increase so how many dimensions need to be considered if we ask me we say generally we go for two dimensions that's why initially as placed in the solution also for two dimensions because two dimensions explaining 86.2% age or 0.8 6 - but third dimension does not contribute too much so we are not considering the third dimension we can go with the two dimensions first dimension and second dimension only two dimensions we are able to see okay now let us come to Rose rose means products now this table explains what is the relationship between the product values that is we have different type of products we see this contribution in this first second dimensions we are taking if you see the first dimension Shampoo is 0.34 3 which is close to zero this next one is 0.276 this is two somewhat closer the distance I am talking about if you talk about second dimension body so is totally separate does not have anything closer to it so here the values expenses how much contribution of points to inertia of dimension more closer the values more they are having relationship good relationship the other they are apart they have very less relationship that is what we can understand from this table the same thing we can see in column values column is fragrances we see rows till lavender same we can understand here also we see sandal which is 0.3 which is close to Jasmine okay if you see the sandal there is nothing close to Jasmine there is nothing close to sandal this and separately okay so this is what we can see the contribution of point two inertia of dimension this two table we're talking about row table and column table can be better understood using the graph you can see this is a two dimension table which is four row points for products can clearly see that here oil and shampoo are having close relationship body lotion deodorant and face cream this is having close relationship and bath soap is totally towards left bottom corner if you see this is two left top corner we have body lotion deodorant and face cream whereas the hair oil and shampoo is in the middle of the right side you can see the point-wise also on dimensions we can observe this so hair oil and shampoo customers are taking as closer whereas this three body lotion deodorant face cream as a group and bath soap as a different group let us see the same thing for fragrances okay what you understand is Jasmine and lily fragrances people treating closer lavender and rose also having better relationship whereas sandal is standing totally apart from other fragrances and menthol is towards positive side towards the right bottom side sandal is towards left bottom that is towards negative scientist and the sandal is towards positive side but what do you understand relationship is that lily and Jasmine are closer and the lab and and Rose are truth sandal is totally apart from other group and the menthol is also a part is a separate now let us see the combination of both nominal variables that is fragrance and product this is what I shown in the beginning of my video in PPT we see it is in a multi-dimensional space hair oil jasmine and shampoo with menthol is having relationship we can see majority of customers said bath so related to sandal fragrance so this can be used by the company to manufacture bath soap my dark go for sandal wood fragments we talked about hair oil they must go with the fibers of jasmine and you can see Body Lotion is having close relations with the lavender and good rent with rose even face cream also customers are taking as a rose this correspondence analysis is useful for having relationship between nominal variables in a multi-dimensional space this correspondence analysis is very useful in research especially in marketing research I hope this video is helpful thank you

Channel: My Easy Statistics

Views: 30,879

Keywords: correspondence analysis, multiple variate analysis, spss, amos, analysis, binomial logistics regression

Id: 42Ts6JoG12A

Length: 18min 30sec (1110 seconds)

Published: Sat Jan 14 2017