Day 2 Team C(Nitin)

BA class as on 4^thsept 2012. By team C

Nitin Dhantole

Types of scales:

Scale is in the range

Ordinal is in an order like greater than and less than

Nominal scale has no order in sequence

In chi square if it is >than 0.05 then null hypothesis is accepted hence which proved that there is no relation in the two variables we have selected as a part of our hypothesis

Data has to be seen locally. And intuition has to be used to analyses a data and it can done with help of experienced person.

In the retail store format done in class.

Employee contact vs stores gives relation when they are in contact but the data shows when the contact increases the satisfaction decreases

With the help of above analysis using chi square we can compare two variables and create a hypothesis

Types of analysis:

Cluster analysis:

1.       Hierarchical (<50)- includes objects like people, brands, etc

2.       K-means (>50) objects are present

Proportionate hazard analysis used in hospitals is a technique used to forecast the room availability. But how can forecast a patient’s death? Predictions can do on this technique by use of regression in this technique.

Medici Effect: book on creativity and innovation.

Divisional cluster:

A single object is divided in to number of objects

Agglomerative cluster:

Multiple objects is compiled to make one object

Clustering process:

1.       Objective is to selection of criteria for the variable to be used

2.       Way to measure distance like people place, etc such as we can use correlation, average in the variables

3.       Thirdly is clustering criteria where two objects are close to each other they can be combined or clustered and the way how this combined cluster can be measured to determine the distance between the other variables

Agglomerative clustering:

A distance matrix is created and two objects of least distance are combined to create a single object.

And till the end all the objects are combined to create a smaller matrix in the end which created only one cluster in the end.

In the end a dendogram is created as an interpretation of all the combined clusters

Distance measurement:

Types are interval, counts, and binary.

Measure dist. from one object to other. It can be a binary measurement, Euclidean distance ( a straight line distance), block distance(a staircase format distance), cosine Chi-square

Clustering criteria:

1.       Where objects are near to each other they can be clustered in to one object called as nearest near criteria.

2.       Distance between the clusters is furthest object.

3.       In centroid clustering we measure distance from center of the objects.