Data Analysis and Insights

The analysis of the data was conducted using SPSS version 23. Therefore, in cases of a numerical analysis it is expected that there could be a variation of +/- 0.02 from other analysis, although such numerical variations are not substantial when interpreting the results of an analysis. As per the data given the number of respondents/ consumers included in the data set include total up to 221. Most of the consumers included in this data are male then followed by female as indicated in table below; there were a total of 132 men (59.7%), and 89 women (40.3%).

The researcher used 130 variables to measure the various outcomes in the designated population. The data was effectively recorded with no missing values. However, there are some data recorded in words(strings), which might present various challenges when analyzing the data. Data such as social media followers (Facebook, twitter and Instagram) are already in numbers and therefore, when analyzing data, such variables should be transformed to numeric so that the analyst can get a clear picture of the whole data. Besides, the variable on connection with others (conn_other_text) is also recorded in strings and therefore cannot be used for analysis. In such a case, the analyst should assign numerical values to the different answers given for this variable to be eligible for further research. Apart from the string values listed above, the data is ready for analysis and does not need any further processing.

The data given can help the analyst in analyzing various relationships that can help in informing a business. The data set has various dependent and independent variables that are commonly preferred by business analysts when analyzing consumer preferences, consuming patterns, and the purchasing power. Dependent variables such as age, gender, education, religion, ethnicity and marital status among others can help in determining various trends in the data.

For example, from a cumulative point of view, most people spend between $50-$99 in their last purchase.

But cumulatively men spent more money in total than women as shown below

Even so, this does not mean that men spend more than women. Earlier we noticed that the number of male respondents were more as compared to their female counterparts. Therefore, when analyzing the which gender has a tendency of spending more than other, using the sum figures shown above is likely to mislead the reader. Instead there is need to conduct a correlational analysis so that we can know the correlation between gender and expenditure. Since men have been ascribed a numerical value of 1 and women 2, a negative correlation would mean that men spend more, while a positive correlation will prove that women spend more.

As shown in the table above gender has a negative correlation with expenditure. Therefore, according to this data men are likely to spend more as compared to women in their last purchases.

In terms of education and expenditure- education levels have been ascribed higher values therefore if there is a positive correlation between education levels and last expenditure, then it would mean that educated people tend to spend more than the less educated one, while if there is a negative correlation between the levels of education and expenditure then it would mean that educated consumers spend lesser than the less educated.

The positive correlation shown above implies that the more educated a consumer is, the more they are likely to spend more. In fact, the data show that when a consumer’s education level increases just by one step, then they are likely to double (.213) their expenditure from the previous rates. From such a small insight it is clear that men spend than women, while educated consumers are likely to spend more than the less educated ones. Therefore, one can conclude that educated male consumers (postgraduate degree holders) are the highest spenders, as compared to other groups.

In grouping the consumers, a K-mean clustering was used, this is due to the fact that K-mean offers tighter clusters as compared to other types of clustering such as hierarchical, especially in the cases have almost similar characteristics. The study used the following variables for clustering, response to sky5 promotion on clothing last month, response to sky5 promotion on accessories last month, response to sky5 promotion on home items last month, response to sky5 free returns on orders last month.

The analysis presented below is a 2-cluster representation that is based on the clients response to the promotions set by sky5 on different products. With the 2 segments, cluster 1 has 91 cases, while cluster 2 has 88 cases. The likelihood to rent clothes in future separates these two clusters

It would also be important to view the results when one uses a 3-cluster solution for the same variables. When clustered in the 3 segments, cluster 1 had 70 cases, cluster 2 had 40 cases, which 69. The intent to rent clothes provided the separation among the three clusters.

In both 2-cluster and 3-cluster analysis all the variables had a cluster of 0.000, which means that the cluster analysis is statistically significant.

The study also used factor analysis using varimax rotation to determine factors that would drive clients to rent clothes. The first factorial analysis was done among the male consumers, while the second one was done among female respondents as highlighted

Continue your journey with our comprehensive guide to Customer Trust and Data Privacy.

Factorial Analysis- Male

Factorial Analysis- Female